Dynamics Center. User Manual. Innovation with Integrity. Version 002 NMR

Transcription

1 Dynamics Center User Manual Version 002 Innovation with Integrity NMR

2 Copyright by Bruker Corporation All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form, or by any means without the prior consent of the publisher. Product names used are trademarks or registered trademarks of their respective holders. This manual was written by Dr. Klaus Peter Neidig September 11, 2017 Bruker Corporation Document Number: P/N: H For further technical assistance for this product, please do not hesitate to contact your nearest BRUKER dealer or contact us directly at: Bruker Corporation Am Silberstreifen Rheinstetten Germany Phone: nmr-support@bruker.de Internet:

3 Contents Contents 1 Introduction Installation Availability Installation from a Bruker Software DVD Installation from the Internet Help License Information General Information on FLEXlm Licenses (Windows, Linux) The License File license.dat (Windows, Linux) Starting the Flexlm License Manager (Windows, Linux) Locating the Correct Host ID (Windows, Linux) Obtaining Flexlm Licenses, Adding them to the License File (Windows, Linux) FLEXlm Version Numbers (Windows, Linux) FLEXlm Diagnostics (Windows, Linux) Licensing on Mac CodeMeter licensing How to Start the Dynamics Center Starting the Dynamics Center on Linux Workstations Starting the Dynamics Center on the Windows PC Starting the Dynamics Center on Mac Starting the Dynamics Center from TopSpin Working with the Dynamics Center General Aspects Working with the Method Tree, Methods & Projects Working with the Built-in File System Tree Drag & Drop Drag & Drop of Spectra to the Display Area Drag & Drop of Spectra to the Method Tree Drag & Drop of a Project File to the Method Tree General Display Features Cross-hair Cursor Direct Display Actions using the Mouse Actions Using the Icon Tool Bar Actions Using Context Sensitive Popup Menus Description of the Spectrum Popup Menu Items Description of the Peak Popup Menu Items Tools Available from the Main Menu Bar File Config Help General Dynamics Method Center H _2_002 iii

4 Contents 6.1 Introduction The T1, T2, T1rho Methods Sample Data Analysis View Report Export The Diffusion Method Sample Data Analysis View Report Export Automated Execution from the Diff Software The Kinetics Method Sample Data Analysis View Report, Export Using 2D data The Cross Polarization Method Sample Data Analysis View, Report, Export The REDOR Method Sample Data Analysis View Report Export The Arrhenius Method Sample Data Analysis View Report, Export Recommended Pulse Programs Protein Dynamics: Basic Relaxation Analysis Introduction Sample Data Analysis iv H _2_002

5 Contents 7.5 View Report Export Further Information Recommended Pulse Programs Summary of Fit Functions Protein Dynamics: Modeling Backbone Dynamics Introduction How to Execute the NOE/T1/T2 Method Extracting Dynamic Information from T1, T2 and NOE Strategy Used with the NOE/T1/T2 Method Performing analysis with the NOE/T1/T2 method Settings TauC Reduced SD Isotropic Modelling Anisotropic Modelling View Results Obtained with the NOE/T1/T2 Method Report results obtained with the NOE/T1/T2 method Export Results Obtained with the NOE/T1/T2 Method Validity of the NOE/T1/T2 Modelling Protein Dynamics: Modelling Side Chain Dynamics Introduction How to Execute the Sidechain Dynamics Method Extracting Dynamic Information from T1 and T1ρ View Results Report, Export Protein Dynamics: NOE BuildUp Analysis Introduction Sample Data Analysis View Report, Export Protein Dynamics: S/N Comparision Introduction Sample Data Analysis View Report, Export Protein Dynamics: CEST Introduction Sample Data H _2_002 v

6 Contents 12.4 Analysis View Report Export ProteinDynamics: Multi T2 / Kd Introduction Sample Data Analysis View Report, Export Protein Dynamics: Interfacing to Relax Time Domain Dynamics Method Center Introduction The TimeDomain1D Method Sample Data Analysis Fit Functions Start Parameters ILT (Inverse Laplace) ILT regions View Context Sensitive Pop-Up Menus Report Export The TimeDomain2D Method Sample Data Analysis View Report Export The qsrc Method Sample Data Analysis View Report Export The qsrc(3) Method Automation Introduction Drag & Drop Command line vi H _2_002

7 Contents 17 InsightMR Introduction Using the Dynamics Center Tutorials Protein Dynamics Tutorial General Dynamics Tutorial Kinetics Tutorial Time Domain Dynamics Tutorial Further Information Contact List of Figures Index H _2_002 vii

8 Contents viii H _2_002

9 Introduction 1 Introduction In general, the Dynamics Center is designed for the analysis of experiments related to NMR data that contain at last one dimension which is not Fourier transformed. It currently consists of three groups of applications. One is called General Dynamics and includes relaxation, diffusion, kinetics, cross polarization and REDOR experiments. The second group is called Protein Dynamics. This part essentially covers the functionality of the former Protein Dynamics Center. Included is the analysis of relaxation, hetero nuclear NOE and relaxation exchange experiments but also the complete protein backbone isotropic and anisotropic modelling based on reduced spectral densities and Lipari Szabo formalism. With the current release the fitting of NOE build up curves and monitoring of protein Sidechain Dynamics using 2 H relaxation have been included. The third group is called Time Domain Dynamics. This part covers the analysis of time domain relaxation data, such as raw FIDs or CPMG decays. For this part.dps files as generated by Bruker's the minispec are used as input files. In general, data can be stored as series of 1D, pseudo 2D, series of 2D or pseudo 3D spectra. Relevant Bruker pulse programs mostly yield pseudo 2D or pseudo 3D spectra. The Dynamics Center recognizes involved pulse program parameters which minimizes necessary user input as compared to data acquired otherwise Figure 1.1: A View of the Dynamics Center with Access to Protein and General Dynamics 1 General Dynamics 2 Protein Dynamics 3 Time Domain Dynamics The analysis of the data is always done in a method oriented sequence, i.e. there is a T 1 method, T 2 method etc. Each method consists of 6 components: Sample Data Analysis View Report H _2_002 9

10 Introduction Export. All methods are accessible in a method tree in which nodes represent the individual methods and leafs the individual components. Figure 1.2: A View of the Method Tree, with the T2 Method Node Opened Some of the steps must to be executed (Sample, Data, Analysis), others are optional (View, Report, Export). The execution order of components is as shown in the tree, first Sample, then Data, then Analysis, etc. A method explicitly applied to a sample and data is called a project. Projects typically include details of calculations, selected parameters and results. Projects can be saved to project files and can be reloaded. The Dynamics Center is a pure Java application. The GUI shows standard components like the method tree, file explorer tree and menu. Embedded is a very powerful multiple object viewer, which is used for all visualization tasks. Furthermore, the Dynamics Center contains a calculation engine (not visible) used for the numerical parts, in/out functions and converters for import and export tasks. Availability and Outlook The development of the Dynamics Center is ongoing. The current version 2.5 contains Protein Dynamics, General Dynamics and Time Domain Dynamics. A further application is online reaction monitoring, and is offered as a separate product named InsightMR. It includes a specially configured Dynamics Center. Organization of this Manual Chapters 3-5 describe general aspects of installation, program start-up, licensing and using the GUI. Chapter 6 describes the various methods contained in the General Dynamics part. Chapters 7-14 are designated to Protein Dynamics. Chapter 15 covers aspects in regard to Time-Domain NMR data. The last two chapters relate to Automation and Tutorials. 10 H _2_002

11 Installation 2 Installation 2.1 Availability As a pure Java application the Dynamics Center runs on all platforms able to execute a java run time environment (jre). Available versions currently exist for Windows XP/Vista/7/8/10, Linux CentOS-5 and Mac (MacOS X or later) systems. The software was compiled using Java 1.8.xxx, thus a Java runtime version jre_1.8_xxx is required and contained in the installation. 2.2 Installation from a Bruker Software DVD The Dynamics Center is a component of the Bruker NMR Software Suite and is thus available (e.g. with TopSpin 3.5 or TopSpin 4) on the Bruker NMR Software DVD. It can be installed independently from other components available on this DVD. It is however recommended to take the Dynamics Center from the Bruker download pages, see next section. 2.3 Installation from the Internet The Dynamics Center can be downloaded from the Bruker Web site, e.g: From there you have to login and navigate further to the required products and platforms. The Dynamics Center is available for Windows, Linux and MAC. All installers contain all needed components: Windows, Linux To install the software, execute the file that was downloaded. You may need administrative rights to do so. An installer pops up which guides you through the installation. The Dynamics Center will be installed in the folder you specify (e.g. C:\Bruker\DynamicsCenter), the registration is performed and an entry in the Start/Programs/Bruker NMR Software (Windows) or Applications/Bruker NMR Software (Linux) menu is created. The installation folder is also stored in an environment variable called DYNAMICS_CENTER which can be used, for example, by other programs to locate the installation. Mac Open the delivered dmg file with the Finder and select the DynamicsCenter2.5.pkg file from there. Follow the installation steps. The target installation folder is pre-defined as DynamicsCenter2.5 and stored in /opt. Tutorials On each of the download pages you may also access several tutorials (protein dynamics, general dynamics, kinetics, time domain dynamics). They include a pdf document with step by step instructions and some training data. See the chapter Tutorials [} 179] for more details. H _2_002 11

12 Installation 2.4 Help Manual, About and License information is available at run-time by clicking on the Help Menu selection. Figure 2.1: The Help Pull-down Menu The manual is in PDF format and can be displayed using a PDF reader. The Adobe Reader is the default reader, but if you prefer to use a different reader you can define an alternative program in Config/Preferences/Default PDF viewer. The manual, release letter and license info are stored in /xxx/docu/ where xxx is the installation directory of the Dynamics Center. Due to Bruker policy a License Agreement is provided. It contains license information for Bruker software products, as well as, any third party products provided by Bruker. 12 H _2_002

13 License Information 3 License Information The Dynamics Center contains three major parts: General Dynamics, Protein Dynamics and Time Domain dynamics. To use the Protein Dynamics a Flexlm license key named PROTEIN_DYNAMIC2 is required. For Time Domain analysis a license key named TIME_DOMAIN_DYNAMICS_1 is needed. 3-months demo licenses are available. Apply for licenses on the Bruker website and specify details in the Comments section if the order number is not available or the name of the license key is not visible on the page. The General Dynamics is free. The Protein Dynamics part requires a license called PROTEIN_DYNAMIC2. The Time Domain Dynamics part requires a license called TIME_DOMAIN_DYNAMICS_1. A new alternative (and the future standard) is CodeMeter licensing, see the description at the end of this chapter. 3.1 General Information on FLEXlm Licenses (Windows, Linux) There are two types of licenses: Node locked licenses are made for individual computers. The licensed software can only run on one PC and is verified by the Dynamics Center. A node locked license is ready for use if it is placed into a particular license file (see below). Floating licenses are network licenses. One (or three) computers in the network are defined as license servers. The floating licenses are bound to the license server(s), but any computer in the network can use the licenses as long as non-consumed licenses are available. If floating licenses are to be used the FLEXlm license manager that controls the licenses must run on the server(s). Usually, the installation of FLEXlm installs and starts a corresponding service. FLEXlm must also be installed on the other computers in the network if the license should be used from these locations. A floating license is ready for use if FLEXlm is installed on all involved computers and the license is placed in a particular license file on all the computers (see below). 3.2 The License File license.dat (Windows, Linux) The license file (license.dat) is a text file with a particular format that is stored at a defined location on the hard disk. Usually this is /usr/local/flexlm/bruker/licenses/license.dat on Linux systems and C:\Flexlm Bruker\licenses\license.dat on Windows systems. A default license file is automatically created when installing FLEXlm, it only needs to be updated if licenses are added to it. Below is an example of a license file: SERVER nike 6906d DAEMON bruker_ls /usr/local/flexlm/bruker FEATURE PROTEIN_DYNAMIC apr \ BBCE384950ACB123476CBE The first line identifies the license server by name and host ID. Lines 2 to N contain individual licenses. Line 1 and 2 must be available, but are not evaluated if only node locked licenses follow. H _2_002 13

14 License Information If you do not want to keep the license file in the default location, but rather somewhere else on disk, you must define an environment variable and set it correspondingly. Example in Linux: Bourne Shell: : LM_LICENSE_FILE=/usr/local/Bruker/licenses/license.dat export LM_LICENSE_FILE or C Shell: setenv LM_LICENSE_FILE "/usr/local/bruker/licenses/license.dat" Example in Windows: Execute Start/Settingd/Control Panel/System and click the Advanced tab. Define LM_LICENSE_FILE among the environment variables. 3.3 Starting the Flexlm License Manager (Windows, Linux) The license manager only needs to run on license server(s) if floating licenses are used. It is usually started during the installation. If needed it can also be started manually: Linux cd /usr/local/flexlm/bruker lmgrd start -c /usr/x/flexlm/bruker/licenses/license.dat If the license manager is already running and you want to stop and restart it, then execute: lmgrd stop -c /usr/x/flexlm/bruker/licenses/license.dat lmgrd start -c /usr/x/flexlm/bruker/licenses/license.dat Windows Use Services in the Control Panel (or in the Administrative Tools with Windows 2000) to stop and restart the FLEXlm License Server. Normally, the license manager will appear with the Startup attribute set to Automatic. Double-clicking on a service allows the service to be started, stopped, or parameters to be defined. Figure 3.1: Listing of Installed System Services 14 H _2_002

15 License Information There is also a tool called lmtools.exe included with the FLEXlm installation. It resides in the FLEXlm installation directory, e.g. c:\flexlm\bruker and offers a graphical user interface. All actions, such as changing settings, re-reading license files, stopping/starting servers etc. can be easily accomplished using this tool. 3.4 Locating the Correct Host ID (Windows, Linux) When node locked licenses are used, the Host ID of the computer where the Dynamics Center will run, must be supplied. When floating licenses are used, the Host ID of the license server(s) is needed. The easiest way to display the correct Host ID is to install the Dynamics Center, start it and check Help/About. Alternatively, you may use system tools, e.g. ipconfig /all on a Windows system, whereas the Host ID shows up under Physical Address. On Linux systems the shell command /etc/sysinfo provides the Host ID. 3.5 Obtaining Flexlm Licenses, Adding them to the License File (Windows, Linux) Licenses are typically sent by together with instructions how to add them into the license file. Generally you can use any text editor to add the licenses to the license file, but when doing this, make sure that you don t change the format of the file. It is particularly important that you don t add special (non-printing) characters, such characters, for example, may be included in the programs. Also note that the licenses (so-called feature lines) are typically very long and line breaks (the \ character) have been inserted for readability. These line breaks must not be removed. When using node locked licenses put the obtained license into the license file on the machine on which the Dynamics Center should run. It is ready to use immediately. When using floating licenses put the obtained license into the license file(s) on the license server(s). The obtained license must also be put into the license file(s) on all computers in the network on which the Dynamics Center will run. The running license manager on the license servers must then be notified that the license files changed. Newer versions of FLEXlm handle this automatically. If for some reason this doesn t seem to happen, execute the lmutil lmrerad tool as described in FLEXlm Diagnostics (Windows, Linux) [} 15]. 3.6 FLEXlm Version Numbers (Windows, Linux) With TOSPIN 2.0 (or later), FLEXlm version 9.2 is delivered. The Dynamics Center is prepared correspondingly. In general, the version of the license server must be larger or equal to the version of the client libraries linked to the Dynamics Center, i.e. must be >= FLEXlm Diagnostics (Windows, Linux) Unfortunately, problems with licenses are quite frequent. A typical example is that users first work with a node locked demo license then buy a floating license, install it and it doesn t work. Analysis of problems in the past years show that the main reasons include: The license file or license string were corrupted when adding a license. Not all relevant license files have been updated. The license manager is not running at all. The license manager is running but didn t see the license file change. Less frequent reasons include: The license string is corrupt as such. H _2_002 15

16 License Information The user supplied the wrong Host ID when requesting the license. There is a conflict in version numbers. The FLEXlm installation folder (e.g. c:\flexlm\bruker) contains a utility tool called lmutil. It can be executed from a command shell and fed with several arguments, for example: lmutil lmreread can be used to inform the license manager when license.dat changed. lmutil lmstat -a display the current status, e.g. which licenses are checked out. lmutil lmdiag provides detailed diagnostic output. When the interactive lmtools program is running on your system you may use this instead. We recommend using Help/About in the Dynamics Center to view the Host ID of the computer, the location of the license file and available licenses. Comparing these values with the output of lmutil lmdiag may reveal the source of the problem. If the problem persists you may send the output of lmutil lmdiag to Bruker via (for e- mail addresses see Help/About) and briefly describe the problem. If the problem can still not be solved Bruker will send back a more detailed diagnostic tool to execute, and send the resulting output to Bruker. Finally, there is also documentation provided with FLEXlm itself, e.g. the document enduser.pdf which is the help folder in the FLEXlm installation directory that can be used for diagnostics. You may also check the Macrovision Web pages, at Licensing on Mac On Mac only node locked licenses are available for the Dynamics Center. The license file is named license.mac located in /Library/Application Support/Bruker/licenses/ by default. A FLEXlm license manager is not needed. The licenses are directly evaluated by the Dynamics Center. As on the other platforms the general dynamics part is free to use if there is already a TopSpin license or a PROTEIN_DYNAMIC2 license available while the protein Dynamics part in any case needs a PROTEIN_DYNAMICS2 license. To obtain a license from Bruker the hostid of the computer must be provided. Either start the Dynamics Center without license and read Help/About to get the hostid or use tools available from the system. Under System Preferences check Network. Under Ethernet / Advanced / Hardware the MAC address is shown. This is equivalent to the hostid. 3.9 CodeMeter licensing A new alternative (and the future standard on all AV NEO spectrometers running TopSpin 4.0 or later) is CodeMeter licensing. The web page for getting CodeMeter licenses is currently under construction. For early usage of CodeMeter again go to and indicate in the comment field that CodeMeter licensing is wanted. The license department will then process the order details and generate an activation ticket. Such a ticket consists of a 25 character code like e.g.: PH3T4-9D9U9-FNSGP-J9FXP-TTNXC This ticket can then be entered in the CodeMeter License Central WebDepot ( which serves to transfer a license to your computer. Internally the package CodeMeter Runtime is required for this. It is automatically installed if TopSpin 4 is installed on your computer and will be available on all platfoms (Windows, Linux and Mac). For further Information please look up the CodeMeter License Management User Manual (see TopSpin 4.0, Help/Manuals). 16 H _2_002

17 How to Start the Dynamics Center 4 How to Start the Dynamics Center The Dynamics Center (DC) is developed and available as a standalone software that is fully compatible to any TopSpin version. After installation of the DC it can be started from desktop icons (Windows, Linux) or Applications (Mac). For simplicity it can also be started from inside TopSpin. 4.1 Starting the Dynamics Center on Linux Workstations On Linux computers click on the program icon Dynamics Center on the desktop, or execute Applications/Bruker NMR Software/DynamicsCenter from the pull-down menu. Figure 4.1: Starting the Dynamics Center via Applications/Bruker TopSpin. Alternatively, one may use an Explorer (or shell window), navigate to the installation folder (e.g. /opt/bruker/dynamicscenter) and execute the start-up script rundynamics. 4.2 Starting the Dynamics Center on the Windows PC On the Windows PC click on the program icon Dynamics Center on the desktop, or execute Start/Programs/Bruker NMR Software/DynamicsCenter from the command prompt. Figure 4.2: Starting the Dynamics Center via Desktop Icon. H _2_002 17

18 How to Start the Dynamics Center Alternatively, one may use the Explorer (or cmd shell), navigate to the installation folder (e.g. c:\bruker\dynamicscenter) and execute the start-up script rundynamics 4.3 Starting the Dynamics Center on Mac Most simply, use the Finder / Application and double click to DynamicsCenter2.2 there. This launches the software and is equivalent to executing the startup script rundynamics located in /opt/dynamicscenter Starting the Dynamics Center from TopSpin In general, it is advised to always take the latest Dynamics Center from the Bruker download pages and use it as a standalone program. If TopSpin is used in parallel the currently loaded spectrum in TopSpin is also known to the Dynamics Center if Settings/Preferences/Default Spectrum Path is selected correspondingly, see next section [} 34]. Furthermore, you can use existing peak lists from TopSpin without doing peak analysis in the DC again, e.g. 1D peak lists that are then projected to pseudo2d spectra or 2D peak lists that are projected to pseudo3d spectra. Complete peak analysis including automated and manual options are available in the Dynamics Center if needed. For routine applications one may also use the Dynamics Center directly from TopSpin, either just by launching it or have a data analysis be prepared in TopSpin and sent over to the Dynamics Center. This is achieved with a special Flow bar option which was originally available for diffusion data but later extended to other experiments. The idea is to offer various steps (processing, peak picking, analysis) in a guided way. Details are described in the Diffusion manual, chapter The flow bar is located in the TopSpin main menu. Figure 4.3: TopSpin Flow Bar for Dynamics Center Support To get the flow bar use Applications/Dynamic/Prepare for Dynamics Center. Currently supported are experiments that yield pseudo2d spectra. Included is a manual peak picking based on a trace of the pseudo2d spectrum. Peaks are stored in a 1D peak list file and the Dynamics Center is notified. It automatically uses these 1D peaks and projects them into the pseudo2d. A direct peak analysis of pseudo2d or pseudo3d data as available in the Dynamics Center can currently not be done in TopSpin with reasonable quality. 18 H _2_002

19 Working with the Dynamics Center 5 Working with the Dynamics Center 5.1 General Aspects The main purpose of the Dynamics Center is to run diverse data analysis in a method oriented way. Different methods (T 1, T 2 and T 1rho etc.) are offered in the method tree shown in the method center. Protein Dynamics, Time Domain Dynamics and General Dynamics each have their own method center. To navigate to the method center, click on the corresponding tabs in the bottom area of the GUI Figure 5.1: GUI Components of the Dynamics Center 1 Main Menu, Icon Toolbar. 3 Graphic Area, Spectra Display. 2 Method Tree. 4 Tabs to get to the method centers or the file system tree. It is possible to switch between the method tree and a file system tree by clicking on the corresponding File tab in the Tab Area. The File tab offers a filtered file tree explorer, whereas only files understood by the Dynamics Center appear. The file explorer may be used when additional spectra not related to a particular analysis method should be occasionally loaded for comparison. The standard loading of spectra is part of executing a method and normally no other option (either from the File pull-down or via the File tab) is required. The main menu bar and icon tool bar are only used for general purposes, such as setting preferences. Only limited entries are offered here since the method tree in combination with spectral popup menus offer all the tools needed. For example in order to analyze T1 relaxation data open the T1 method in the Method Center with a left mouse button click and execute the Sample, Data, Analysis, View, Report, Export components one after the other. H _2_002 19

20 Working with the Dynamics Center Most of the entries in the Settings/Preferences pulldown menu are used to customize the display, e.g. colors. A useful option relates to spectra selection. Users that work with the Dynamics Center and TopSpin in parallel often want the dataset that is currently display in TopSpin is selected by default if a new method is opened in the Dynamics Center. This can be achieved be setting the Spectrum Default Path. Figure 5.2: Setting the Spectrum Default Path If the option Get latest spectrum from TopSpin is chosen the corresponding name of the dataset is automatically inserted in Data if a new method is opened. The TopSpin installation path must be specified. The settings remain active even if the program is closed. In principle it is possible to call the Dynamics Center directly from within TopSpin and get a lot of information passed automatically. Since the implementation is not yet complete and it is not guaranteed that the most recent Dynamics Center is used we recommend to use the Dynamics Center in its standalone form. A second option to mention is Default Slider Type. Most of the methods use series of spectra or pseudo spectra. Whenever data are loaded a slider window is supplied. It allows you to navigate through spectra, traces or planes. Figure 5.3: Standard Data Slider The slider comes in a separate window that is shown in foreground. You can move it around and the latest position will be memorized and used the next time. You can also iconify it if is in in the way. Nevertheless many user dislike to have these additional windows. Therefore you can change the Default Slider Type to Add slider to toolbar if space permits as done in the following example: 20 H _2_002

21 Working with the Dynamics Center Figure 5.4: Sliders Added to the Main Tool Bar Two sliders could be added to the icon tool bar and used to navigate through the data. They are however fixed in position and size and cannot be iconified. If available space is consumed the default slider windows will be used. 5.2 Working with the Method Tree, Methods & Projects To open the method tree click on the Protein Dynamics, Time Domain Dynamics or General Dynamics tab. Usually, one is already open when the program starts. The method tree operates like a regular file tree explorer, to open/close a method node move the cursor to the +/- fields and click the left mouse button. After opening a node the components Sample, Data, Analysis, View, Report and Export are visible. When these are activated with a left mouse button click, dialog windows are displayed or calculations are made. All methods contain the same components, but are customized for their respective requirements. Figure 5.5: Left and Right Click to the Method Tree A method that is applied to a sample and its data may be called a project and can be saved to a project file. This is a xml file and may contain a large amount of information. It is not intended to be read by the user, but is used to save the status of an analysis and later load it H _2_002 21

22 Working with the Dynamics Center again, e.g. to continue the analysis or to review the results. Actions that can be performed on projects are available through popup menus. To access the popup menu of a method, move the cursor to the name of the method node and click the right mouse button. The following actions are currently available: Open A project file is opened that was previously saved using the Save command. By first clicking on one or more of the components Sample, Data, etc. on the method tree you can open the exact settings that have been saved for the project. It is possible to work with several methods/projects at the same time, e.g. a T 1 and T 2 project. It is also possible to have two or more projects open for the same method, e.g. several T 1 projects. The method tree must first be extended, refer to the Add to Tree command below. When executing open project file of one method, e.g. T 2, it is possible to specify a project file name of a different method e.g. T 1. This is useful if the sample information is the same and one does not want to input this information again. Details in the Data, Analysis, etc. components are typically different and would have to be checked/corrected when traversing the method tree. Instead of executing open you may also just drag & drop a project file (created via Save, Save As) to the method name. The method is executed as far as described in the project file. Finally, you may also start from the system file explorer and drag & drop spectra or folders to a method. The usability of spectra is checked and a selection list is shown in unclear cases. Execution of Sample and Data is then performed. Warning: Using project files across applications areas, e.g. open a T 1 project created in general dynamics in T 1 of protein dynamics is not recommended because the supported types of spectra is different, i.e. general dynamics may use pseudo3d, 2D, pseudo2d and 2D spectra while protein dynamics usually only offers pseudo3d and 2D spectra. Save A project can be saved to a specified file using this command. File names and extension may be freely selected, the default extension is.project. We recommend saving different project files in a folder of your choice, for example c:\tmp\ on Windows systems. A typical example of a project file name would then be c:\tmp\t1.project. Save As This command can be used to save a project to a different file. Close This command is used to close a project and clean up memory. To save the project before closing the Save or Save As commands may be executed before. Suspend Working with projects usually includes the display of data and results. Often three display windows are used per method. When working with several methods in parallel the number of display windows increases and the size of each window becomes smaller and smaller. If the windows of a particular method are not really needed on the display one can suspend the project. All data and results are kept in memory but the display windows disappear until Resume is executed. 22 H _2_002

23 Working with the Dynamics Center Resume Resume re-displays the display window closed when Suspend is used. Add to Tree Use this command to add another method of the current type to the tree. For example, if executed from the popup of a T1 method, another T1 method appears on the tree. Its name will be T1(1) and T1(2), etc. if applied again. Only the parent method (e.g. T1) offers Add to Tree. The duplicated methods (e.g. T1(1)) only offer Remove for being removed if no longer needed. Remove Used to remove duplicate methods if no longer needed. General Remarks The contents contained in the project file and what is re-calculated at run-time is determined by the software. Part of the stored information is related to spectra names which are currently kept with absolute path names. If a project file and data are copied to another computer it is advised to maintain the spectra names. The contents of the popup menu are context sensitive. For example, when a project is loaded and nothing has been changed, Save is inactive, or if a project is suspended only Resume is active. 5.3 Working with the Built-in File System Tree Each method contains a component named Data which is used to load spectra and peaks. Occasionally, independent other data may be dragged and dropped to the display area for reasons of comparison. Either the regular desktop explorer (e.g. Windows Explorer) or the built-in file system tree can be used. Figure 5.6: First Tab on the Left: File System Tree. To open the built-in file system tree click on the File tab. A file system tree opens but only shows folders which contain spectra, optionally molecules and peak lists. A popup menu to configure the tree is displayed with a right mouse button click on the tree. H _2_002 23

24 Working with the Dynamics Center Figure 5.7: File System Tree with Popup Menu. Set file system root This command can be used to change the root of the tree. For example, when all data is stored in d:\data and any corresponding sub-folders, you can set the root to d:\data. The tree performance is then faster as it contains less components. Select file type Spectra (1r, 2rr, 3rrr) are automatically contained in the tree, this command allows selection of Mol files (.mol, sdf) and/or peak lists (.xml,.peaks). 24 H _2_002

25 Working with the Dynamics Center 5.4 Drag & Drop Each method contains a component named Data which is used to load spectra and peaks. For display purposes or to speed up an analysis various drag & drop options exist Drag & Drop of Spectra to the Display Area To get spectra displayed they may be dragged and dropped from the internal file explorer or from the system desktop explorer. The resulting display depends on the target position of the drop and on the current display status. Figure 5.8: Drag & Drop Target Positions. 1. The display area is empty After drop the display is filled, no matter if the drop position is near A, B or C 2. If the display already contains a spectrum a drop to A generates a small overlay display in the upper left region, a drop to B generates a full overlay display and a drop to C splits the display area and creates a new internal window 3. When generating overlay displays restrictions apply related to dataset dimension and spectral ranges. If an overlay cannot be generated a new display window is launched instead. 4. 3D spectra may be dropped as well. In this case a dialog window pops up and allows you to select the type of display (which plane or which projection). If planes are selected a slider is shown which allows you to navigate to other planes. H _2_002 25

26 Working with the Dynamics Center Drag & Drop of Spectra to the Method Tree Each method contains a component named Data which is used to load spectra and peaks. Spectra or folders may be dragged and dropped from the desktop explorer directly to a method in the method center. If possible the method will be started and executed up to Data. For example, a single pseudo2d spectrum containing mixing times in a vdlist can be dropped to the T 1 method or a series of 1D spectra can be dropped to the kinetics method. Folders containing sub-folders and spectra may also be dropped. If they contain different types of spectra and the software does not know what to do, all spectra are shown in a list for selection. Figure 5.9: Drag & Drop a Series of 1D Spectra to the Kinetics Method. If you drag and drop a folder that also contains a suitable xml file describing a complete analysis then this xml file will be executed and applied to all possible spectra. For further information see chapter Automation [} 171] Drag & Drop of a Project File to the Method Tree It is also possible to drag a project file from the desktop explorer and drop it on a method name (e.g. T 1, T 2, etc.) of the method tree. Method names are highlighted while moving the cursor to indicate the current drop position. After the drop the method is completely executed (Sample, Data, Analysis, etc.) according to the settings contained in the project file. If the project file is dropped to a method other than described in the project file, a warning is displayed. 26 H _2_002

27 Working with the Dynamics Center Recently more flexibility has been added to make drag & drop more convenient. As shown in the following picture, not only drop on a method (A) is possible but also drop to the root of the method Figure 5.10: Options to Drop a Project File tree (B, Dynamics or Method Center) and finally also on the display area (C) on the right which is normally used for data and result display. As also pointed out in chapter Automation [} 171], each time a project file is generated (Save, or SaveAs from the popup after right click on method) a corresponding xml version of the project file is also generated. This xml can be dragged and dropped the very same way, in fact we recommend to use the xml files instead of the projects files as the xml technology is newer and will offer more options in future. 5.5 General Display Features The display is generated by a powerful multiple object viewer which can handle internal and external windows, different types of overlays and geometry layouts. Many features are not necessarily needed when doing a method oriented analysis, but the user may want to use them in specific cases, e.g. if he wants to drag & drop another spectrum that is not part of the analysis into the display Cross-hair Cursor Whenever the mouse pointer is moved into the spectrum display it turns into a cross-hair cursor. When moving this cross-hair, the corresponding positions (e.g. in ppm units) are displayed in the upper left part of the display. If several internal spectral display windows are visible the cross-hair automatically appears as a linked cursor if possible. H _2_002 27

28 Working with the Dynamics Center Figure 5.11: Linked Cross-hair Cursor. In the above example the linked cross-hair cursor is shown in two of the spectra windows Direct Display Actions using the Mouse The most frequent activities using the mouse are: Zoom into a spectral region of interest. Click left button --- drag --- release. Overview (full spectrum) Double click left button (anywhere on spectrum). Change vertical 1D scale or contour levels Turn mouse wheel. If the display window contains several internal spectral windows, one of them is always active. Define one internal display window as active Single left button click to internal window. 28 H _2_002

29 Working with the Dynamics Center Actions Using the Icon Tool Bar The icon tool bar contains a set of icons used to change the display. Due to the availability of direct actions, the context sensitive popup menus and the methods available in the method tree, tool bar icons are generally not needed for display setup. Therefore, it is possible to configure the icon tool bar via Config/Preferences/Select individual icons from the main menu bar and keep for example only needed icons. Figure 5.12: The Icon Tool Bar. The usage of the icons should be self-explanatory, as the tool tip text is displayed when the cursor is moved over an icon, thus only short explanations are given here. OPEN When the open icon is selected, a standard file open dialog is launched. By navigating down to a spectrum, e.g. e:\data\guest\nmr\xxx\1\pdata\1\2rr the spectrum is loaded. More elegantly you can also use drag & drop from the built-in or desktop file explorer. Open is not used to select spectra for analysis since this is done via the Data component offered by each method. The purpose of Open is rather to load other data for reasons of comparison. PRINT When the print icon is selected a PDF copy of the currently visible objects is generated and stored to a specified file. To get suitable output of relaxation results, use the Report component on the method tree. INPUT OPTION The input options icon is used to access a dialog allowing selection of internal and external windows, and different types of overlays, for the multiple object viewer used to generate displays. Data loaded via Open, Open TopSpin also follow this selection. Drag & Drop has its own rules as described in the previous section. The relaxation methods also have some built-in rules to guarantee proper displays. GEOMETRY Geometry is used to select different types of layout geometries for the multiple object viewer. CORRELATION The zoom correlation, when active, will zoom in on the selected object and will also recalculate the display of other objects if possible. If for example a 1D and a 2D spectrum are on screen, zooming into the 1D spectrum will also change the horizontal spectral region of the 2D spectrum. RESET The reset icon is used to restore the default vertical scale values if they have been changed. H _2_002 29

30 Working with the Dynamics Center VERTICAL OFFSET The vertical offset icon is used to change the vertical offset. To do so: 1. Click on the icon and move the cursor to the desired object, e.g. a 1D spectrum. 2. Click on the desired object and move the mouse up or down while holding the mouse button down. 3. Release the button when you have reached the desired position. In case of 2D contour displays constant values are added or subtracted to all contour levels. VERTICAL RESET The vertical reset is used to restore the vertical offset to the original values. ZOOM To zoom into a region of interest, move the mouse to a corner of the region, click the left mouse button - drag the mouse - release when finished. Since this is a frequently used operation, it can be used on spectral objects without first clicking on the zoom icon. FULL DISPLAY This icon is used to open a full object display (e.g. full spectrum, complete histogram, etc.). Since this such a frequently used operation you only need to double-click with the left mouse button on an object. UNDO This icon is used to switch back from the currently selected display region to the previously selected one. CONTOUR LEVELS This icon opens a dialog window used to set contour levels. Special functions like automatic ramp filling or Bezier-smoothing can then be used to improve displays. For quick vertical scaling you only need to turn the mouse wheel Actions Using Context Sensitive Popup Menus Context sensitive popup menus are displayed by clicking the right mouse button on an object. The content and behavior of the popup menus are context sensitive, which means that there are different popup menus and behavior depending on the cursor position (e.g. near a peak or not near a peak). 30 H _2_002

31 Working with the Dynamics Center Figure 5.13: Context Sensitive Popup Menus Description of the Spectrum Popup Menu Items When the right mouse button is clicked in an empty spectral region, a popup menu appears. The available entries depend on the type of spectrum and current display. It is for example possible to display a set of 2D spectra in 3D mode or a set of 1D spectra as a stacked plot. The popup menu then changes accordingly. Most of the popup menu entries are self-explanatory, some of them are briefly described. Toggle full display The toggle full display is used to toggle to and from full screen display. This is often needed if several display windows are shown in the viewer and individual objects become quite small. All other objects remain in memory and re-appear if toggle is reactivated. Visibility The visibility command displays a list of all objects that can be selected individually for display. With some objects it may be possible that the internal logistics does not allow the current visibility to be changed. H _2_002 31

32 Working with the Dynamics Center Add peak This command adds a peak at the cursor position where the right mouse button was clicked. If not peaks but peak integration regions are used for the further analysis (see description of Data in the next chapter) the popup menu displays Add peak integration area instead. In some applications one frequently wants to define several regions at once. Another entry named Add multiple peak integration areas is then shown. Configure update style If a peak is added at the current position in the current spectrum it may automatically be added in other spectra or not. This is controlled by the update style. Measure distance The measure distance command is used to measure the distance between two points. To measure a distance click the left mouse button on the starting point, drag the mouse to the end point and release the mouse button. The distance between the starting point and current cursor position is shown. Set contour levels With 2D or pseudo 3D spectra this command opens a dialog window used to set contour levels. Special functions like automatic ramp filling or Bezier-smoothing can then be used to improve displays. For quick vertical scaling you only need to turn the mouse wheel. A graphical version of this command named Graphical contour levels shows a color pyramid with two handles that allow you to set the lowest and highest levels with the mouse. Properties This command displays some properties of the current object Description of the Peak Popup Menu Items The data analysis in the Dynamics Center is based on peaks which can be picked in various ways. Intensities or peak integrals are then extracted. In some cases users prefer to work with regions (using region integrals) instead of peaks. Automated and interactive options exist to define such regions. Depending on which choice has been made (peak or region) the text in the peak popup menu refers to peak or peak integration area. When the right mouse button is clicked near a peak (the peak is highlighted in this case) a popup with options related to peaks appears: Annotate This command is used to assign a name to a peak. There is no limitation in length of the name. Normally, peak names would refer to the measured nuclei, e.g. HN in a hetero nuclear 1 H 15 N HSQC spectrum or CH 3 in a 1D spectrum of a small molecule. If, in case of protein spectra, the residue number should be included, like HN10 or HN-10 or HN/10 or HN [10] etc., other useful displays, e.g. T 1 relaxation times vs. residue number can then be generated. Depending on the update style (see below) the annotation can automatically be transferred to the other planes/traces/spectra. Some methods, like Kinetics, also allow you to specify the number of nuclei (e.g. number of protons) corresponding to a signal. The purpose of this is to use the number for normalization. The number of nuclei is also put into the peak name. 32 H _2_002

33 Working with the Dynamics Center Add A new peak gets created at the position where the cursor currently is. If using peak integration areas an area is defined with a default size. Since one often wants to define several areas, Add multiple is also available. Each area needs to be defined with the cursor: Click to one end of the area, drag to the other and release mouse button. You can define area by area until you click the Stop button in the help window that pops up while the command is active. Move With this command a peak or region can be moved with the mouse. Click the left button, drag the peak to the desired position and release. Delete This command deletes a peak when the mouse is right clicked on it. Delete in a region All the peaks inside a selected region can be deleted with this command. To define a region, left click on a starting point, drag the mouse to the end position and release. Keep in a region If offered, this command keeps all the peaks inside a selected region, all others are deleted. To define a region, left click on a starting point, drag the mouse to the end position and release. Delete diagonal peaks in planes In case of homo nuclear pseudo 3D spectra, e.g. 3D COSY-i-DOSY one often wants to perform analysis with cross-peaks only. In this case all diagonal peaks can be removed. Search by name If offered, use this command to search for a peak containing a given string as part of the peak name. The purpose is to easily locate individual peaks in crowded displays. An automated zoom into the corresponding spectral region takes place. Undo last peak change This is used to undo the most recent peak manipulation. Configure update style If you manipulate peaks (annotate, add, delete, etc.) you usually want that the corresponding action is done in all other spectra, planes or traces as well. If you add a peak in the first spectrum all other spectra get a peak at the identical position. This is the default behavior. In case an automated update of other data is not wanted apply Configure update style. Save peak list to disk All peak manipulations are done in the computer memory. The peak list on disk is not changed until a save peak list to disk is applied. When a spectrum or project is closed, it is checked whether peaks have been saved already or not and you are asked what to do. H _2_002 33

34 Working with the Dynamics Center Be aware that the peak list storage should be consistent with the peak options selected under Data/Peaks (see below). If you had for example selected automatic picking there and then improved the peaks manually and finally saved the peaks to disk they could be lost the next time you work with the data. This is automatically checked and the peak picking option is silently changed to avoid loss. Properties This command displays some properties of the current peak. 5.6 Tools Available from the Main Menu Bar Tools provided in the main menu bar are not needed for the actual data analysis but mainly serve for customization. The Config pull-down menu for example, contains several entries to change display settings, fonts etc. The Dynamics Center 2.5. release comes with a new look and feel as compared to earlier versions. Commands in the main menu bar are now shown as icons. Figure 5.14: Dynamics Center Before Version 2.5 Figure 5.15: Dynamics Center 2.5 Throughout this manual the text often refers to the GUI style before version File Open, Open TopSpin 1D, 2D, 3D These commands can be used to open and display corresponding data but are mainly available for historical reasons. Especially, data loaded this way do not participate in any of the methods shown on the method tree but are rather displayed for comparison. Close multiple objects The standard workflow is to load data via Data available in each method and continue with analysis and viewing results, see below. When a method is closed all its data and display windows are also closed. If you also have added other data to the display, e.g. via the internal file explorer or by drag & drop from the desktop explorer each corresponding display 34 H _2_002

35 Working with the Dynamics Center windows offers close on their popup menus. Using that gets cumbersome if many display windows need to be closed. In this case it is faster to use the Close multiple objects command. Properties Displays the properties of the currently loaded objects. Print Creates a PDF copy of the currently visible objects. Copy to clipboard Copies the currently visible objects into the system clipboard. Exit Exit is used to close the Dynamics Center. If needed, the projects should be saved using Save on the method tree before using the exit command Config Input Options The input options command is used to access a dialog allowing selection of internal and external windows, and different types of overlays, for the multiple object viewer used to generate displays. Data loaded via Open, Open TopSpin also follow this selection. Drag & Drop has its own rules as described in the previous section. The relaxation methods also have some built-in rules to guarantee proper displays. Geometry Geometry is used to select different types of layout geometries for the multiple object viewer. Scaling Opens a dialog offering several options for vertical scaling spectra when spectral are shown in overlay. Best results are obtained when different spectra contain a reference peak each to which the data can be scaled. Display Options Offers various display options, e.g. if spectra names should be displayed. Arrange Arrange allows you to change the display without closing and re-loading objects. There are however limitations, only objects of the same type can participate. For example, you can t overlay a histogram to a 2D spectrum. Correlation/Zoom The zoom correlation, when active, will zoom in on the selected object as well as all other objects if possible. H _2_002 35

36 Working with the Dynamics Center Preferences Preferences offer a number of options are offered to customize the display: Different sets of icons for the tool bar. Less than 20 icons are offered to customize the display and only a very few are actually needed in daily work. The option Select individual icons allows you to define which icons are to be shown at all. The selection is maintained over different program runs. Font options. Color options. The Default PDF Viewer can be selected. Default Spectrum Path can be set. For typical offline usage and not having TopSpin installed you can define a folder that is used as a starting point for spectra selection in each method. If Topspin is also used there exists a so-called current data set which is most recently used by TopSpin. In many cases this is the spectrum that shall be analyzed next in the Dynamics Center. Thus, an option exists to use it automatically for spectra selection. Since more than one TopSpin can be installed on the computer, you must define the installation path of TopSpin to be used. Default Output Folder can be set. It is presented as a default in several dialog windows. User Interface Two possible user interface styles are offered. A Windows-based user interface called Windows Look and Feel. A platform independent user interface called Cross Platform Look and Feel. Figure 5.16: Available User Interface Types. The Native user interface should show the user interface compatible to the current platform, e.g. on Windows XP systems it would look different compared to Linux. The platform independent user interface creates a user interface that is platform independent and looks the same on all systems. Changing the user interface within a running program can cause update problems therefore you have to confirm that all display objects are first closed. You have to reload them afterwards. Regardless of the type of interface you can decide if the internal windows come with decorations or not. The advantage of decorations are that they contain buttons that provide quick access to internal windows, however decorations take up a lot of space. 36 H _2_002

37 Working with the Dynamics Center Help Manual The Adobe Reader (Windows, Linux) or Preview (Mac OS) is launched to display this manual which is stored in PDF format in the docu folder inside the Dynamics Center installation folder. If no Reader is found an error message appears and you will need to define an alternative PDF viewer using Config/Preferences/Default PDF Viewer. On Mac one can for example just specify open there which internally calls the default reader, e.g. PreView. About The about command displays a short summary of names, version numbers, license status and contact address. License info Displays a list of several license information files. The contents of the files can be displayed by selecting with the left mouse button. H _2_002 37

38 Working with the Dynamics Center 38 H _2_002

39 General Dynamics Method Center 6 General Dynamics Method Center 6.1 Introduction The General Dynamics method center is used for the analysis of T 1, T 2, T 1rho, Diffusion, Kinetics, Cross Polarization and REDOR data through the selection of corresponding methods. Each of these methods contains the same components, Sample, Data, Analysis, View, Report, and Export. Their components have to be executed in this order. After a successful execution the corresponding nodes are green color-coded on the method tree. A red colorcode indicates that a component is currently executing, e.g. doing a longer calculation. Any subsequent components can only be executed if the previous component is shown in green. Figure 6.1: Color Codes indicate the State of a Method In the example above state A shows the Diffusion method just opened. State B indicates that Diffusion is opened and Sample and Data have been successfully executed, while Analysis is currently in progress. The subsequent component (View) can only be executed after analysis has turned green. However, it is possible to execute components of other methods in parallel. The Sample component is almost identical for all methods but other components, especially Data, Analysis and View are customized to cover typical application scenarios. For example, Data of the T 1, T 2 and T 1rho methods allow the use of pseudo 3D, series of 2D, pseudo 2D, or series of 1D spectra, while the Cross Polarization method can be used for pseudo 2D, or series of 1D, but REDOR only for pseudo 2D spectra. 6.2 The T1, T2, T1rho Methods With this group of methods T 1, T 2 or T 1rho relaxation times can be determined. In case of T 1 three different experiments are commonly performed: Exponential decay, saturation recovery and inversion recovery. Bruker T 1 pulse programs often used include t1ir, t1irpg, cphirt1, cphsatrect1, cpht1.av, cpxt1, cpxt1.av, satrecechot1, satrect1, satrect1.av and satrecechot1.av. T 2 data are typically acquired with the pulse programs cpmg, cpmg1d, cpmg1dpr, and T 1rho with cpht1rho, cpht1rho.av, cpxt1rho and cpxt1rho.av, and zght1rho. H _2_002 39

40 General Dynamics Method Center Sample The Sample component is used to provide information about the sample and is only used for report purposes Data The Data component is used to define details related to the spectra and peaks. Groups of parameters are accessible via different tabs. Spectrum tab Used to select the type of spectrum, available options are pseudo 3D, series of 2D, pseudo 2D and series of 1D. Many Bruker pulse programs yield pseudo 3D or pseudo 2D data. Pseudo data must contain the relaxation dimension in F1. The advantage of using these pulse programs is that parameters and variables (mixing times) involved are automatically available. In case of series of 2D or 1D, necessary parameters must be entered manually. Selecting options is context sensitive as can be seen by enabled or disabled dialog fields. Figure 6.2: Data Dialog/Spectra Tab for the Selection of Spectra Peaks tab Allows selection of peak picking techniques. All analysis methods are based on peak intensities/integrals. Peak picking and peak integral calculation are therefore very important. The automated picking has a built-in adaptive peak picking threshold estimation. Noise levels are determined for each data point. The most important source of information is that the spectra are available N times (N planes or traces in pseudo spectra, or N individual spectra). The signal-to-noise in these planes/traces/spectra differ based on selected experimental parameters (e.g. mixing time) but real peaks should occur in the majority of the planes/traces/spectra. So-called epsilon values define local search radii and are used to decide if peaks in different spectra are really different or not. If a peak picking has been done before and saved (see save peak list to disk in the peak popup menu) a peak list file named peaklist.xml exists in the folder where the spectrum (1r, 2rr, 3rrr) is. It has the same name and format as peak lists generated by TopSpin. The second peak option will load such an existing peak list at spectrum. 40 H _2_002

41 General Dynamics Method Center Using a peak list from another spectrum makes sense if there is a well picked reference spectrum whose peaks can be used in other spectra. An interesting feature is that such a peak list may have the same but also a lower dimension. For example a good 2D HSQC peak list can be used in all planes of a pseudo3d spectrum or a good 1D peak list can be used for all traces of a pseudo2d spectrum. The Dynamics Center does all necessary calculations automatically. Since there may exist slight peak shifts between imported peaks and current spectrum so-called peak snapping is usually applied. This means that imported peaks are moved to near experimental local maxima when possible. One snapping algorithm combines a global shift and local nearest neighbor search. This works well if the global shift can be automatically detected looking at some outlying peaks. In many cases the more simple snap using a local neighbor search gives better results provided the snap radius does not need to be selected so large that snapping to neighboring peaks becomes a problem. In cases where peak positions are very stable along different spectra (planes, traces) it is even better to do snapping in the first spectrum only, then just copy peaks to the others. Identical digital resolution and calibration of all spectra is required. If snapping is applied the specified epsilon values (see lower part of the dialog window) are used. If given in units of data points typical values for pseudo3d/2d are 3-5 points and 5-10 points for pseudo2d/1d data. The automated peak picking sometimes picks too many peak and an interactive threshold based picking is preferred. In case of 1D or pseudo 2D spectra a horizontal bar appears which must be moved up or down with the mouse while the left mouse button is clicked. Finish positioning the bar by releasing the mouse button. Peaks above this threshold are picked in the first spectrum and then searched for in the others. In case of 2D or pseudo 3D spectra you must define a signal free region with the mouse. Left click on one corner of the region, then drag to the other and release the mouse button. The peak picking then used peak picking thresholds derived from these definitions. In case of pseudo 2D data each column of the corresponding 2rr can be formally regarded as a peak provided it contains significant intensity. This option is called use all columns above threshold and is only offered because of historical reasons. Again a horizontal bar appears which must be moved up or down with the mouse while the left mouse button is pressed. Finish positioning the bar by releasing the mouse button. The number of columns defined this way is usually much higher than the number of peaks picked normally, and subsequent calculation times increase correspondingly. The second last option is called do manual peak picking later and means that you want to do the peak picking yourself. A reason is typically that the spectra are so simple that a few needed peaks are easily defined by hand. Once peaks are available the text of this option changes to keep currently available peaks. It means that the current peaks should not be overwritten by another picking. This situation frequently occurs if for example only the integration parameters change. Example: A peak picking was done, peak intensities were selected, analysis was performed. Now you want to see if the analysis changes if you had used peak areas instead of peak intensities. You go back to Data/Integrals and change this option. But you want to keep all peaks as they are. Therefore make sure that under Data/Peaks the option keep current peaks is selected. Finally, in case of pseudo3d data only, there is an option named advanced peak analysis. It uses construction/re-construction techniques based on singular value decomposition (SVD) to find a suitable peak picking threshold for each data point. In many cases the results are very satisfactory but also intensive in computing times. It is important to know that at any time existing peaks can be modified. Right click to the spectrum to get a popup with an option to add peaks. If at a certain cursor position this is activated a new peak appears exactly at that position. You can also right click near a peak to get a special peak popup menu which contains further options (delete, move, rename etc.) The interactive behavior and options depend on the selected type of integrals, se next section. H _2_002 41

42 General Dynamics Method Center Attention: After manually changing peaks execute Save peak list to disk from the peak popup menu. The peak picking option is automatically switched to just keep available peaks. In case of typical kinetics experiments the sample changes due to chemical reactions. This may cause large peak shifts that cannot be captured with a snapping technique. Therefore the kinetics method offers a tic box to activate peak tracking. This peak tracking works best if integral regions instead of peak intensities are chosen, see next section. Integrals tab This tab is used to select peak integrals. It is often good to just take the peak intensities, or as an alternative peak areas (user defined) can be used. This makes sense if peak shapes are bad and a simple region is more valid. If not predefined for the given spectrum, each peak is equipped with a default region and shown on the display. These regions can be moved/resized interactively using a mouse. Move the cursor to the center, or border areas of a region. The color will turn red, click the left mouse button and drag to the desired position. Click the right mouse button on a region to display a popup menu that offers some useful tools for deleting and adding regions or for setting all regions to a given size. 1 Figure 6.3: User Defined Peak Areas: Move/resize with Cursor 1 Cursor was moved into region that turned red. While the left mouse button is clicked the region can be moved/resized. Regions are automatically stored in the spectrum folder whenever peaks are stored. In fact peaks and regions are always in a one-to-one relation. If peaks change, e.g. get deleted, corresponding changes are done to the regions. This is different from earlier releases of the Dynamics Center which allowed peaks and regions to be independent of each other. Integration by line shape analysis is an automated alternative that is suitable when peaks are well isolated and peak positions are identical to peak maxima. This is the case after automated peak picking. Peak shapes are detected during a later stage of the analysis and 42 H _2_002

43 General Dynamics Method Center are not shown on the display. Finally, peak de-convolution can be used, which is based on the currently picked peaks. A careful (manually optimized) peak picking is therefore advised. Different shapes (Gaussian, Lorentzian, mixed) can be applied. Suitable start parameters for the peak line widths should be given. The spectra popup menu contains measure distance as a graphical tool to estimate peak widths. The de-convolution algorithm used can apply correlation of parameters of different peaks, indicated by slow and fast calculation mode. Even with fast mode, minutes are typically needed for a full pseudo 3D spectrum with 12 planes and about 100 peaks each with Gaussian shapes, which is recommended as default. The variable Gaussian/Lorentzian shape needs 2-3 times longer. You must wait until this is finished before continuing with Analysis of the current method. However, you may work with other methods in parallel. Illegal operations, e.g. closing the method while the de-convolution is calculating, are blocked during this time. It is possible to cancel the peak de-convolution, typically it takes a few seconds until the calculations are stopped. Lists tab Various aspects of the lists of variables (e.g. mixing times) are handled in a context sensitive way. In case of T1, T1rho experiments run with Bruker pulse programs like t1ir, t1irpg, cphirt1, cphsatrect1, hpht1.av, cpxt1, cpxt1.av, satrecechot1, satrect1, satrect1.av, satrecechot1.av a vdlist exists which contains the mixing times. When acquiring T2 data with Bruker pulse programs, there is no vdlist but a vclist which first must be converted into a vdlist. This conversion is done automatically and uses the number of mixing times, loop duration and constant duration which are read from the acquisition parameters. For example, when using hsqct2etf3gpsi3d the loop duration is equal to D31 but when using trt2etf3gpsi3d or cpmg it is calculated from P2 and D20. If the values cannot correctly be obtained they must be entered manually. The question remaining is how to treat repetition experiments. It is recommended to measure the one or other mixing time (or whatever the variable is) more than once to check for reproducibility of the data. Repeated experiments can be kept or collapsed into mean data. Repetition experiments are also evaluated when calculating uncertainties of peak intensities or peak integrals. Currently, 2 different options are offered to estimate systematic errors. The estimate can be based on variance averaging using peak intensities/integrals. All repetition experiments and peaks are taken into account, and the square root of the average variance is then assigned to all the peaks at each mixing time. The other alternative is to use all repetition experiments to calculate the largest difference of peak intensities/integrals per peak and assign this as a systematic error per peak for each mixing times. Variance averaging balances but sometimes underestimates systematic errors whereas the difference method slightly overestimates systematic errors. TD tab If available (pseudo 2D or pseudo 3D spectra), used to select the Time Domain (TD) relevant parameter. It is initially set to TD effective as originally defined in TopSpin. This indicates how many planes or traces are available. It may now be reduced to smaller numbers, e.g. if there are just empty planes or traces. H _2_002 43

44 General Dynamics Method Center Figure 6.4: Data Display with Slider Finally, the specified data are loaded and partially displayed. A data-slider which allows you to select other planes/traces/spectra appears. The slider can be regarded as a player which contains buttons forward/backward play, fast forward, fast backward and stop. Another button offers additional settings to be selected. When several methods are active with several data displays open then several sliders will be shown, one slider for each method. These sliders normally work independently of each other but can also be correlated, see options offered via the settings button. Any data-slider may be closed if not needed (especially if several sliders are displayed). Deleted slides can be generated again re-executing Data. If multiple sliders are displayed they obtain automated positions to avoid full overlap. Additional Data Display features Apart from inspecting individual spectra by using the slider it is often helpful to look at all data simultaneously. An overlay of all 1D spectra (or all traces of a pseudo 2D spectrum) shows, for example, if signal positions are preserved. This is important when defining signal integration regions manually. Likewise a stacked display of 1D spectra shows how signal intensities change. The Perspective and height can be adjusted by the user to get a proper display. In case of pseudo 3D or series of 2D spectra a true 3D display which can be scaled, rotated and moved helps to get a quick overview over all peaks. For an example check the Diffusion method in section The Diffusion Method [} 54]. The various display commands can be executed from the spectrum popup menu. 44 H _2_002

45 General Dynamics Method Center Figure 6.5: The Popup Menu of 1D and pseudo 2D Spectra offers Stacked Plot Display For example, after selecting stacked plot on/off from the spectrum popup the following display is obtained: Figure 6.6: Stacked Plot Display of a Series of 1D Spectra Below the method center area two sliders are activated to change the perspective and height of the display. Further options are available from a popup menu of the stacked plot. The stacked plot is a regular display object, standard operations like scale up/down with the mouse wheel or zoom in/out are possible. Peaks of the first spectrum or trace are displayed Analysis The Analysis component is used to fit the peak intensities/integrals to particular functions and extract the relevant relaxation parameters. It can be executed if the Data component was already successfully executed before. H _2_002 45

46 General Dynamics Method Center Figure 6.7: Example of an Analysis Dialog Window, T1 Method The upper part of the window relates to the selection of the fit function. Available options depend on the selected method. For example, in a T2 experiment a simple exponential decay is given by 46 H _2_002

47 General Dynamics Method Center It is also possible to fit a constant offset to the selected function in which case the equation is: Adding an offset should however only be done if a real offset is expected. If the measurements only stopped at some point and the decay has not yet reached 0 then this is an incomplete decay rather than a complete decay with offset. The fitted T2 values would be different! The number of components can be set to values greater than 1 if a peak is overlapped and more than one parameter (relaxation time) can be extracted. For example, setting the number of components to 2 would fit the bi-exponential function with 2 amplitudes and 2 decay constants. The number of components is understood as the maximum number of components. If for example 3 are chosen an internal test fit with 1, or 2 or 3 components is performed and compared via the AIC criterium. Only the best solutions is presented. If one wants to see if a fit with the specified number of components was possible, even if statistically not the best solution, one can activate the tic box prioritize highest number of components. In general, the maximum number of components is defined by the number of mixing times (e.g. 16) divided by the number of fit parameters per component (e.g. 2 or 3) but fitting more than 3-4 components often yields limited results. The non-linear Marquardt algorithm is used to perform the curve fitting. Start parameters are either determined automatically or can be defined by the user. An alternative to the given fit functions, offset term and number of components, would be to define your own function in Python style. Any function is given in a defined notation. The function value must be called y, the variable is x and fit parameters are given as an array called p. There may be constants involved if entered as numbers. An example would be which would correspond to where I, T and C are fitted. Comma separated start parameter values must be given in this case, for example: 1.0e7, 5.5, 0. The amplitude, here p[0], must be provided in numbers as found for intensities in the spectrum, e.g. 1.0e7. While moving the cursor in a 1D or pseudo 2D spectra the intensity is shown. Properties of peaks obtained via the peak popup menu show peak intensities. In 2D spectra the contour level suggests a typical intensity. It is usually sufficient to estimate the right order of magnitude of the amplitude. Later on, when displaying or reporting fit results (see View, Report below), p[1] is regarded as the most relevant parameter. As fit parameter unit any comma separated strings can be provided, if there is no unit, specify none, e.g. none, s, none in the above example. H _2_002 47

48 General Dynamics Method Center Three options related to error estimation are offered: Use the Y data (peak intensities, integrals) under the assumption that their uncertainties are unknown but equal for all Y values. The non-linear fit determines errors for the fitted parameters from the inverse of the un-weighted curvature matrix (second derivatives of chi-squared). The final chi-squared itself is arbitrary. This option would be useful if the error in the data is obviously larger than estimated by signal-to-noise calculations or repetition experiments. The second option applies if individual uncertainties of the Y values are known and are passed to the non-linear fit. The errors of the fitted parameters are then calculated from the inverse of the weighted curvature matrix. The uncertainties themselves are derived from the standard deviation of the noise in each plane/trace/spectrum and, if available, differences of Y values in repetition experiments. Use a Monte Carlo simulation. The non-linear fit is performed 500 times with the input Y data varied according to a normal distribution, with a standard deviation equal to the uncertainties of the Y values. The second or third options are preferred. If the resulting errors of the fitted parameters seem to be too small than option 1 is an alternative. In order to provide fitted parameters within a confidence interval a confidence level needs to be given. The confidence interval of a fitted parameter is then calculated by multiplying the error of the fitted parameter with a factor taken from the inverse of Student-s-T cumulative distribution at given confidence level and number of degrees of freedom. The confidence interval calculates as fitted parameter +/- fitted parameter error. The larger the chosen confidence value, the larger is the interval. Note: Fit results and errors are usually similar for the different error estimation method used. In some cases fits may fail or not depending on the selection. This depends on the data quality and on the errors found for the peak integrals/intensities. Sometimes these errors appear unusually small. An alternative to fitting a selected number of function components is the application of an inverse Laplace transform (ILT). This technique is commonly used in the analysis of diffusion data but can be applied to relaxation data as well, currently however only to exponential decay, saturation recovery and full inversion recovery and only when pseudo 2D or series of 1D spectra have been selected. The key of ILT is to explain the data by a linear combination of terms evaluated at fixed points. The user only needs to carefully specify the expected range of relaxation times. The minimum and maximum expected numbers should be extended a bit (e.g. 20%) to avoid numerical artifacts during the calculation. For other settings the default values are usually good to use, especially grid type = log grid, regularization = second derivative, find alpha automatically. The grid optimization tic box can be tried, it sometimes yields better solutions. For the interested user mathematical details can be found in S. Provencher, Computer Physics Communications, 27, , (1982). 48 H _2_002

49 General Dynamics Method Center Figure 6.8: ILT Tab of the Analysis Dialog Window IILT is computational expensive, can be instable and yields larger errors of fitted parameters. However, it finds possible numbers of components by itself and is the method of choice if distributions of relaxation times are expected. If the data yield individual relaxation times then regular fitting with 1-3 components is faster and better to apply. H _2_002 49

50 General Dynamics Method Center View The View component allows viewing of the obtained results. Important is to relate spectra/ peaks with parameter fits shown as individual fit curves. The user may select which types of displays to view. Figure 6.9: View Dialog Window to customize the Result Display Show fit curve in separate internal window means that a new display window is created in the main window and filled with fit curves. In order to get fit curve displays on screen you must navigate the cursor to a peak. Depending on the selected option if cursor is moved to peak or if mouse button is clicked at peak the corresponding curve pops up immediately or needs a left button click. When moving to the next peak the fit curve is replaced unless the cumulate fit curves tic box is switched on. In that case one selected curve after the other is overlaid. Duplicate curves are avoided. Additional global cumulate opens a further display window and adds the fit curves there as well. This additional window is also open to other projects and thus allows to add different fit curves of different data into the same window. Display residuals shows the difference between experimental and fitted data. These differences should be randomly distributed. A statistical 2-sided Shapiro Wilk test is available via properties of that plot. If the cumulated fit curve tic box is switched on residuals are also cumulated. If the selected fit function permits it, a logarithmic y axis can be selected. Further display details are available under the Details tab. Error bars (uncertainties, see previous section) can for example be drawn. The fit curve display is a regular display object and can be scaled, zoomed etc. as usual. The X variables (mixing times in the vdlist) may be in arbitrary order and correspond to the order of measured planes/traces/spectra. The fitting procedures need peak integrals in sorted order which is handled automatically. To see which plane/trace/spectrum relates to which point in the fit curves, a green marker is shown unless fit curves with different colors and symbols are shown in overlay. This marker jumps from point to point if you switch to other planes/traces/spectra using the spectra slider. 50 H _2_002

51 General Dynamics Method Center Fit display objects As mentioned the fit display objects are regular display objects and can be scaled, zoomed etc. The fit displays also contain context sensitive popup menus. Figure 6.10: Fit Display Objects also have Context sensitive Popup Menus. Experimental points (black), continuous fit curve (blue line) and possibly obtained points from ILT (red crosses) can be shown. Context sensitive popup menus are available after a right mouse button click. If the cursor was near a point in the fit curve (it will be highlighted with a red color) the popup menu offers: Delete point The current point is deleted and the fit curve is recalculated. Undo latest delete The point deleted last is restored and the fit curve is recalculated. This can be repeated until all previously deleted points are restored. Delete point apply to all A point in a fit curve is often deleted if the data of that particular mixing time are bad. In this case it makes sense to delete this point in all fit curves. Note, that there is no global undo that would restore a point in all curves. Delete point in range All points in a range defined with the cursor (left click, drag, release) are deleted. With undo latest delete you can restore deleted points one by one. Delete point in range, apply to all All points in a range defined with the cursor (left click, drag, release) are deleted in all fit curves of all peaks. If a fit curve is manipulated this way the fit calculation is repeated internally and the display is updated. H _2_002 51

52 General Dynamics Method Center Manipulation of fit curves should only be done in special cases, for example, if one wants to see if a better result can be obtained after deleting bad points. Manipulations of the fit curves are indicated among the properties of the fit curve, but are lost when the program closes. A different situation is that one acquires decaying relaxation curves and later sees that the last measurement points only yield noise. Instead of deleting such points from the fit curves one can lower the TD value in Data/TD (pseudo 2D or pseudo 3D spectra) or chose a smaller number of spectra in Data/Spectra (series of 1D or 2D spectra). When the right mouse button is clicked outside any fit point (no red highlight visible) a popup menu with the following options is offered: Toggle This is the standard display toggle used to show the fit curve in full screen. Visibility Controls the visibility of individual display objects. This is only used occasionally since there are several other ways to customize the display (Toggle, View, Suspend on the method tree). Undo latest delete point The point deleted latest gets restored and the fit curve is recalculated. This can be repeated until all previously deleted points are restored. Add external data This special option allows you to read a (csv) text file that contains comma separated x variable values and y values (without units) in each line, like: , , , These are loaded and displayed together with the fit curve display on screen. Individual color and line style can be selected. The main intention of this tool is to compare, for example, simulated data with experimental data. The relative y-scaling is preserved, in order to get proper displays the external supplier of the csv file should scale the data properly. The range of x variable values should essentially correspond to the range on display. Figure 6.11: Loading external Data, e.g. simulated REDOR Curves. 52 H _2_002

53 General Dynamics Method Center Delete curves Whenever more than one fit curve is shown in a window individual curves can be deleted from the display. Colors & Symbols If different fit curves are overlaid in cumulative mode this additional entry appears in the popup menu and allows you to set colors and symbols for each curve. Properties Some properties of the current fit curve are shown. These include the used x and y values, fit results, goodness of fit and some peak information. Close Closes the fit display, however it will reappear as soon as you move the cursor to another peak in the spectrum or another histogram item if visible on screen. To permanently switch off the fit curve display of a method re-execute the View component and chose the proper setting there. If pseudo 2D or series or 1D spectra have been selected the View dialog window also contains a tab named 2D plot. This serves to display all results similar to DOSY plots known from the diffusion method described in The Diffusion Method [} 54]. The 2D plots show all used spectral peak positions along F2, while F1 is the relaxation axis. The fitted relaxation times and relaxation times errors are translated into F1 positions and F1 widths to obtain the desired 2D plot. If ILT was also used as an alternative to fitting it is possible to get a 2D plot from the ILT results as well. Figure 6.12: Results of a T1rho Analysis with 2D Plots from Fit (upper right) and ILT (lower left). The above picture shows the large fit parameter errors from ILT (as expected) but the overall agreement between fit and ILT is quite good. This can also be seen on the fit curve display where experimental (black), fitted (blue) and ILT (red) are contained. H _2_002 53

54 General Dynamics Method Center Report A PDF report that contains loaded information, as well as results, can be generated in PDF format. Graphical components include the current spectrum and fit curve displays. Numerical components include the sample, peak and fit information. The user may select desired components in a dialog window. The name of the PDF file must also be specified. At the end of the report generation AcroRead is automatically launched to display the report. Some versions of AcroRead do not display the PDF file if it was not specified with its absolute path name, e.g. c:\tmp\test.pdf. If you don t have AcroRead on your computer an alternative PDF display program may selected under Config/Preferences/Default PDF Viewer Export Information, especially peak information and fit results, can be exported to a file on disk. The supported formats are text and xls (or xlsx). This allows users to present data with standard tools, e.g. chart diagrams in EXCEL or to load peak integrals into other software packages to repeat the fit calculations, or fit other functions not offered in the Dynamics Center. 6.3 The Diffusion Method Pulsed Field Gradient (PFG) Diffusion experiments are used to measure random, translational motion of molecules. This provides information about the diffusion process of a certain species, about the environment the molecule is diffusing in, about size distributions of molecules, or to separate different molecules in a mixture by their different diffusion coefficients due to of their different molecular weight. Typically a pulse sequence as shown is applied. Figure 6.13: Possible Pulse Sequence to measure Diffusion Constants. A series of spectra is run where either the gradient strength or little delta (gradient pulse length) or big delta (diffusion time) is changed systematically. In the case of non-restricted diffusion, i.e. liquid state samples, little delta and big delta are kept constant in order to avoid signal attenuation caused by the spin-lattice relaxation time T 1 or the spin-spin relaxation time T 2. The signal attenuation characterized by a decay constant D is then given by where x stands for the variable that was changed during the experiment. Formally, the signal decay can be described by 54 H _2_002

55 General Dynamics Method Center where is called the B value. In this form the change of peak integrals follows a simple exponential decay Sample The Sample component is identical for all methods in the general dynamics method center and serves for input of information that is used for reporting purposes Data The Data component is very similar for all methods in the general dynamics method center but is customized respectively. Figure 6.14: Data Dialog of the Diffusion Method. Diffusion spectra can be of type pseudo 3D, series of 2D, pseudo 2D or series of 1D. The pseudo 3D i-dosy data contain two spectroscopic dimensions F2, F3 and one gradient strength or time dimension in F1. This corresponds to the situation found with pseudo 3D relaxation data. If spectra are measured/processed in a different way, they cannot directly be used, but can still be split into a series of 2D spectra and then used in the Dynamics Center. Contact Bruker if support is needed in this case. Most experiments are however performed with pseudo 2D or series of 1D experiments. The Data component of the Diffusion method differs from the previously described relaxation methods (see section Data [} 40]). Firstly, there is an additional tab named Scaling. If experiments are run in a special mode every two consecutive spectra need to be divided, like spectrum (previous)/spectrum (current) point by point and yield half the number of scaled spectra. This is often used in experiments where big delta is varied and serves to minimize the influence of T1 relaxation to the data. The Scaling tab is only active for pseudo 2D or series of 1D data. H _2_002 55

56 General Dynamics Method Center Secondly, there is an additional tab named DiffusionPar, which displays the diffusion parameters as loaded from the t1par parameter file, which is stored in the same folder as the spectrum. You can modify these values and they will be used in a subsequent curve fitting. They can also be stored into a project file if the diffusion method is stored to disk (see Save As on the popup menu of any method in the method tree). The next time the project is loaded the modified values will be available. However, the t1par on disk which was generated by the diff software (which is part of the TopSpin installation), is not modified. After closing the Data dialog the first trace or plane or spectrum is shown and a slider appears that allows you to navigate to other traces, planes or spectra in the following example a pseudo 3D COSY i-dosy spectrum (M. Nilsson, A. M. Gil, I. Delgadillo, G. A. Morris, Chem. Commun. 2005, ) was loaded and the first plane is shown. Figure 6.15: The 3D Display can be selected from the Popup Menu of a 2D Display. It is even possible to get a 3D display as shown below, use the spectrum popup menu and select 3D display. After selection the 3D display is calculated and shown in addition to the 2D display. The slider used to navigate to different planes also works on the 3D object and indicates the current plane. To rotate the 3D object, click and drag with the left mouse button. The rotation axis is perpendicular to the mouse movement. Double click performs a reset, the mouse wheel is used to scale the size. Further commands are available in the 3D popup menu. 56 H _2_002

57 General Dynamics Method Center Figure 6.16: Example of Pseudo 3D i-dosy Data Set with COSY Spectra in F2, F Analysis The Analysis component is similar to any other method. It serves to select the details of the data modelling. The available fit functions are those that correspond to the various types of diffusion experiments, i.e. if g, δ or Δ is varied. These are complemented by two function alternatives based on B values, one assuming that one or a few specific diffusion constants can be fit to the data, the other assuming that the data is better modelled by a distribution of diffusion coefficients. Like in the analysis of relaxation data, there is tab named ILT, which serves to set up an inverse Laplace Transform calculation as an alternative to curve fitting. A standard software used in this field is CONTIN written by S. Provencher. His publication in Computer Physics Communications, 27, , (1982) describes a lot of details of the underlying mathematics. The Dynamics Center contains its own implementation written in Java and based on the non linear least squares technique NNLS, see for example, the book Solving Least Squares Problems by C. L. Lawson and R. J. Hanson. For a better understanding of the results it is useful to recall some ILT details and compare them to curve fitting. As an example and without any mathematics let us assume there are some exponentially decaying data depending on two diffusion constants. Leaving aside all other details including units. Curve fitting models the data to a given function with a defined number of parameters. We should set up 2 exponentials and expect 2 diffusion constants. The obtained fit result is of arbitrary accuracy and the obtained values can be very close to the true values, say 1.825e-9 and 5.55e-9. ILT models the data to a linear combination of functions, e.g. 10. The number of fit parameters does not need to be known, only the upper limit, here 10. Furthermore, the expected value range of the parameters must be given, say 1.0e-10 to 1.0e-8. A grid with 10 diffusion constants is then put on this value range, for simplicity say 1.0e-10, 3.0e-10, 5e-10, 7e-10, 9e-10, 1e-9., 3e-9, 5e-9, 7e-9 and 9e-9. ILT tries to find weights for these diffusion constants such that if put into the linear combinations of exponentials the decaying data including given constraints are explained as good as possible. Since there is no grid point at 1.825e-9 or 5.55 e-9 these 2 values can never be determined, however H _2_002 57

58 General Dynamics Method Center a distribution of some neighboring values might explain the data. So for example, ILT could find a1*e-9, a2*3e-9, a3*5e-9 and a4*7e-9 (where a1..a4 are some amplitude values) as a possible solution. If the expected number of function terms is known, then curve fitting is the method of choice. If it is unknown, or if there is even a natural distribution of fit parameters, say a distribution of slightly different diffusion constants due to different molecular species, then ILT is the method of choice. Curve fitting is quite fast, ILT is computational expensive. Curve fitting yields a solution that is as close to the data as possible. Neglecting trivial constraints (e.g. diffusion constants are not negative), constraints play a minor role. ILT yields a solution that is close to the data and close to constraints at the same time, the one at the expense of the other. A typical constraint is smoothness of the solution. A socalled regularization parameter determines how important the constraints are considered. Curve fitting depends on the start parameters but are usually not very critical. ILT depends on the choice of grid points, expected value range and the regularization parameter, all of which are very critical. Error calculation in curve fitting is straightforward, e.g. based on covariance analysis or Monte Carlo techniques. In case of ILT error determination is much harder and is, for example, based on the back prediction of individual results. Errors are also related to the choice of the grid. Curve fitting is relatively insensitive to errors of the input data (peak integrals/intensities) in most cases. ILT is extremely sensitive to errors in the input data, small changes can result in very different solutions. The ILT tab shown in the analysis dialog window allows you to tune the ILT calculation: Figure 6.17: ILT Tab of the Analysis Dialog of the Diffusion Method. Minimum and maximum expected diffusion constants define the possible value range spanned by the grid. Avoid choosing too small of a range, at least in an initial calculation. The reason is that ILT can behave unexpectedly if a true diffusion constant is very close to the border of the value range. It depends on how smoothness is calculated. Include offset fit should only be activated if there is a true offset in the data. If a decay has not reached zero because not enough measurement points exist, this is not a true offset. Grid types gaussian, linear or logarithmic are available. If diffusion constants are expected over different order of magnitude chose logarithmic. If the value range is relatively small, gaussian may be a good choice. Linear is offered for reasons of completeness. The regularization technique is either based on derivatives (first and second) or on the norm of the solution vector (Tikhonov). Thikonov is simple and fast, the second derivative computational expensive, but often gives better results. Tikhonov often requires a somewhat stronger regularization parameter. 58 H _2_002

59 General Dynamics Method Center At next, the strength of the regularization must be defined. This parameter is commonly called alpha. Very low values, e.g.e-10, means that only a weak regularization is done, the data might be explained very well, but the solution may contain spurious diffusion constants that are obviously wrong. You can set the value manually e.g. to 1e-3 or leave it up to the software to find a good alpha. The internal criterion used for this is based on reduced alpha values. The automatic alpha calculation requires a substantial number of iterations and increases computing time. The number of grid points usually equals the number of available spectra (or traces, planes). Only if this number is large (e.g. > 100) one may consider to choose smaller numbers as computing times would otherwise increase enormously and the quality of results would often decrease (too many variables). Finally, the underlying linear equations may be reduced by compressing the so-called kernel matrix. This is done by applying singular value decomposition (SVD) and keeping only the most significant eigenvalues. This number can be specified explicitly or given a 0, it will then be automatically determined. Unfortunately, there is no way to optimize all parameters simultaneously and in a simple way. When displaying results, the decay curves as back calculated from ILT, are shown together with the regular fit curve. This gives a good impression how well ILT performed. It should be mentioned that internal calculations are performed in double precision arithmetic. ILT uses a kernel matrix that contains exponential terms which may, depending on the experimental setup, approach zero. To avoid instabilities in the remaining ILT, critical data points at the end of the experimental decays are just stripped off. The following shows a very good ILT example. Expected diffusion constants were around 1e-10, thus the grid was spanned from 1e-11 to 1e-9. The grid points are distributed logarithmically, second derivative regularization with automated alpha determination was used. Figure 6.18: ILT with Log Grid, (middle) Compared to Fit (left). With the same parameter selection, but regularization with Tikhonov and a very small regularization parameter, the obtained result is slightly worse. H _2_002 59

60 General Dynamics Method Center Figure 6.19: As Previous Figure but with Small Alpha and Tikhonov Regularization. Finally, a result is shown where the grid was spanned over a too narrow range. Peaks in the ILT Dosy plot start to shift and the fit curve displays clearly shows that the ILT back calculation (red crosses) does not fit well to the data (black) or fit (blue). Figure 6.20: Problematic ILT Result, Some Peaks Shifted and Back Calculation Deviates from Input. In general the errors of the diffusion constants are large if calculated from ILT as compared to fit. This is indicated as the line width in F1 (axis of diffusion constant). See also 2 Data [} 40] 60 H _2_002

61 General Dynamics Method Center View The View component is used to set the details for viewing results, and is similar to View for any other method. One difference is that the user might want to see the fit curves as a function of different variables. If for example, big delta (Δ) was changed, the fit curve can be displayed as a function δ or as a function of the B value, B = -γ 2 g 2 δ 2 (Δ-δ/3). Similarly, if little delta (δ) was varied the fit curve can be displayed as a function of δ or B. Finally, if the gradient strength (g) was varied the fit curve can be displayed as a function of g, B or the socalled q values defined as q = (-γgδ)/(2π). There is also a new tab named DOSY plot. Figure 6.21: DOSY Plot Tab of the View Dialog of the Diffusion Method. Originally, DOSY plots are 2-dimensional plots in which diffusion constants of peaks are plotted versus peak positions. When displaying a DOSY plot the spectral dimension (F2, horizontally) is derived from the spectrum display. To save computer memory one may chose a smaller number of data points. The display range in F1 (diffusion dimension) must be given by the user. This range should be selected such that the calculated diffusion constants are within this region. We recommend selecting the display range identical to the grid range spanned during ILT, especially if ILT was used. The number of points along F1 often correspond to the number of traces (spectra) of the pseudo 2D (series of 1D). Since this number is often small it can be increased to get a better display resolution. However, the errors of the calculated diffusion constant are scaled as well. An extreme increase of the number of points should be avoided, especially if ILT was involved, since peaks might be split artificially in F1. H _2_002 61

62 General Dynamics Method Center Figure 6.22: Typical Diffusion Result View with DOSY Plots based on Fit and ILT. The diffusion axis can be shown on logarithmic scale which makes sense if the diffusion constants span more than order of magnitude. To better show very large and very small peaks in the same plot the intensities can also be scaled logarithmically. The actual 2D display also depends on the selected contour levels which very often are also logarithmically spaced. Finally the line width of peaks along F1 can be set to a minimum number of data points. The reason is that this line width is based on the error of the fitted diffusion constant. This error may very small and then let peaks appear very narrow along F1. In most cases, displays generated by ILT already have very broad peaks along F1. As the cursor is moved to peaks (or the left mouse button is clicked at peaks) the individual fit curve displays are generated. With a right mouse button click one can get the properties of the fit curve. If ILT was calculated the corresponding ILT results are also part of the properties. The viewing of results based on pseudo 3D i-dosy spectra is very similar. The DOSY object that is calculated is also a 3D object having 2 spectroscopic and one diffusion axis. By default the first plane if it shown together with a slider to navigate to other planes. Peak positions indicate the calculated diffusion constants the peak line widths are derived from the error calculations based on fit and/or ILT. From the popup menu of the plane display you can select a 3D display calculation and for example get the following picture: 62 H _2_002

63 General Dynamics Method Center Figure 6.23: 3D DOSY Display selected from the Popup Menu of the 2D Display. The 3D display object is fully active, if you use the slider to navigate through different planes, the 3D display also changes and indicates the current plane position. If peaks are shown the one in the current plane get a different color. There is a popup menu available on the 3D display as well. It can be used to customize the display. Rotating the display is possible by a left click and dragging the mouse. The rotation axis is perpendicular to the mouse movement. Turning the mouse wheel scales the object up and down. A double left click resets the display to the initial default Report The Report component is similar to the report for any other method and is self-explanatory. If the check box to obtain spectral plots is activated, it will also contain the DOSY plot according to the settings as selected on screen Export The Export component is similar to export for any other method and self-explanatory. If an ILT calculation has been performed the ILT parameters and results will also be contained. H _2_002 63

64 General Dynamics Method Center Automated Execution from the Diff Software The diff software is part of the TopSpin installation and is involved in the set-up diffusion experiment. It also creates the t1par parameter file in the spectrum folder. The Dynamics Center retrieves several parameters from this file. In addition diff also creates an xml interface file named TopSpinDC.xml which is also stored in the spectrum folder. This interface file describes a complete diffusion analysis of a spectrum. The details of the automated diffusion analysis can be set up in diff. Figure 6.24: Dynamics Center Interface in the Diff Software to Set Up and Launch a Complete Analysis. 64 H _2_002

65 General Dynamics Method Center 6.4 The Kinetics Method The kinetics method follows the descriptions given so far, but there are a number of issues to be considered. These are related to the complexity and characteristics of typical kinetics spectra. For example, in contrary to relaxation data, spectral peaks may change position and line shape, or even disappear or build up. Doing a simple peak picking and monitoring peak intensities as a function of a time are not sufficient. The analysis of individual peaks may differ from peak to peak. Thus, it is not possible to only select a particular fit function and apply it to all signals. Examples of references for using NMR in kinetics are: Clegg, I.M., Gordon C.M., Smith D.S., Alzaga R. & Codina A., NMR reaction monitoring during the development of an active pharmaceutical ingredient, Anal Methods, 4, 1498 (2012). Susanne F., Smith D. & Codina A., Kinetic understanding using NMR reaction profiling, Org Process Res Dev., 16, 61 (2012) Sample The Sample component can be used to specify some simple information about the given sample. It is only used for documentation purposes and may be contained in reports or exports Data The current release of the Dynamics Center supports the analysis of series of 2D, pseudo 2D and series of 1D kinetics spectra. A quick way to get spectra loaded is to use drag & drop as described in section Drag & Drop of Spectra to the Method Tree [} 26]. The normal way is to use the Data dialog where the user can provide the spectra names and associated time values given in seconds. Since the number of spectra may be large (512 spectra can be simultaneously handled by the Dynamics Center) an option named auto fill names/values from file is provided to load spectra names and time values from a text file. 1 2 Figure 6.25: Data Dialog of the Kinetics Method with the Option to load from Text File 1 Option: load names & times from text file 2 Option: create a text file here H _2_002 65

66 General Dynamics Method Center The text file looks like: e:\data\dynamic\nmr\kinetics\1\pdata\1\1r 10.0 e:\data\dynamic\nmr\kinetics\2\pdata\1\1r 50.0 e:\data\dynamic\nmr\kinetics\5\pdata\1\1r Each line contains a full spectrum name, a tab separation and a time value (in seconds). Otherwise, there are no format restrictions and it also doesn t matter whether / or \ are contained in the names. Such a file can be generated elsewhere and specified in the next field. It may also be generated in the Dynamics Center by clicking on the details button. However, it is assumed that all spectra are stored under different expno numbers (see TopSpin spectrum name convention) in increasing order corresponding to increasing time values. When activating the details button a further dialog window is presented. Figure 6.26: Simple Tool to create a Text File with Spectra Names and Time Values. To create a new text file select Create and use new list, then select the spectrum name corresponding to the first time value. Starting from this spectrum further spectra names are searched under different expno numbers. There are two ways to assign time values to the spectra. One calculates values from time=time0 + n*δt. Time0 is the delay between starting the reaction and beginning the first acquisition. The value n refers to the n-th spectrum and Δt to the constant time interval between 2 consecutive acquired spectra. The first time (time0), e.g. 10, and Δt, e.g. 60 must be specified. The time values calculated would then be 10, 10+60, , , etc. A second option evaluates a time stamp that is stored together with each spectrum. It corresponds to the start of each spectrum acquisition. The difference between each two consecutive time stamps is used to assign a time value to each spectrum. The values for the first spectrum must be provided by the user. 66 H _2_002

67 General Dynamics Method Center Finally, specify a text file to which the generated list is stored. The next time needed it can then just be loaded from there. Once the dialog window of the list generator is closed (by pressing OK) the list is generated and displayed for inspection. If individual spectra are known to have limited data quality they may be de-activated at this point. After closing the list, the main Data dialog is automatically updated. This includes the number of spectra found. This number can still be changed to smaller values by the user, for example, if it turns out that the last experiments did not provide any good data. The Peaks tab of the Data dialog window is similar to the other methods. However, two items are worth mentioning, manual peak analysis and peak tracking. Figure 6.27: Peaks Tab of the Data Dialog of the Kinetic Method, Top Part. In kinetics, when spectra contain complex multiplets, one may prefer to do manual peak picking or region definition instead of using automatic or semi-automatic analysis (which is sufficient for most other methods). To perform complete manual picking select the option do manual peak picking later. When continuing to work, this option is later renamed into just keep currently available peaks. Further peak picking, region definition or other manipulations are possible at any time. If spectra contain multiplets rather than simple peaks it is furthermore advised to select use peak area (user defined) integrals found under the Integrals tab. Manual work is then based on defining peak integration areas. Figure 6.28: Integrals Tab of the Data Dialog of the Kinetic Method, Top Part. H _2_002 67

68 General Dynamics Method Center Once the spectra are loaded and are displayed on the screen, right click on the spectrum. To manually define peak integration areas chose add peak integration area or add multiple integration areas from the popup menu. Figure 6.29: The Spectrum Popup Menu contains Tools to add Integration Areas. Either a single area (with a default size) or multiple areas can be defined. The latter is preferred as you can adjust regions carefully. Figure 6.30: Adding multiple Integration Areas using the Mouse. 68 H _2_002

69 General Dynamics Method Center If Add multiple peak integration areas is selected, a help window appears which indicates how to define areas. Left click to one end, drag to the other end and release. Repeat for each area and finally click to the stop button when finished. The definition of areas can be done in any of the spectra of your choice or even if all spectra are shown in overlay (see spectrum popup menu). The areas are automatically transferred into the other spectra. The result may then look like: Figure 6.31: Updated Display after having added Integration Areas. Each area has an additional marker in the middle which is referred to as peak. Peaks and peak integration areas are always in 1-to-1 correspondence. When moving the cursor close to an area or peak the context sensitive peak popup menu becomes available. It offers further options like annotate, move, delete etc. As indicated in section Data [} 40] you may also move the cursor into an integration area, where it is highlighted in red color, and move or resize it with the left mouse button. Another major difference to other methods is indicated by a tic box named apply special peak tracking at the bottom of the Peaks tab. Figure 6.32: Bottom Part of the Peaks Tab of the Data Dialog Window. H _2_002 69

70 General Dynamics Method Center Peaks of different spectra have to be related to each other if they belong to the same nucleus. Then intensity (or area integral) changes can be analyzed as a function of time (or other variable). In most cases (e.g. relaxation data) the composition and conditions of a sample do not change from experiment to experiment and peak positions are conserved within a small epsilon radius. This epsilon can be specified, e.g. 5 data points, and is used to search for corresponding peaks. The value should not be too large to avoid false search results. In kinetics the sample conditions may change and peaks can systematically shift, disappear, or appear. A simple nearest neighbor search with a given epsilon is no longer sufficient. Instead, the special peak tracking can be switched on. The details of the tracking depend on the case, e.g. if peaks or peak areas are used or if peaks are defined automatically or manually. The defined epsilon values and/or sizes of defined peak integration areas are taken into account. Systematic changes of positions and intensities are detected by special matching techniques. In general the epsilon should be selected in the order of several typical line widths. The tracker is computational more expensive but is able to follow shifting peaks in many cases. Manual work is needed, especially if peaks are not visible in the first spectrum, but show up later. The reason is, for example, that the automated or threshold based peak picking start in the first spectrum, then pick and track in the others. Signals that do not yet exist in the first spectrum will often not be correctly handled. However, if you add peaks or regions manually you can first navigate to proper spectra using the slider and define peak areas there. The tracking then combines forward and backward tracking to find related peaks in all spectra. Figure 6.33: Shifting Signals of Acetic Anhydrite (left) and Acetic Acid (right). The above example shows two signals with their integration regions defined. Over the series of spectra the signal shift is so large that signals may even cross if displayed in overlay. The integration regions have been defined in one spectrum and were then tracked in all others. Note: There will be cases where the automated tracking fails, e.g. stops or follows a wrong peak. You should then open the peak popup menu by right clicking on a peak and changing the Update style such that further manual changes are only done in the current spectrum. Then use the spectrum slider to go to different spectra and modify peaks or integration areas as needed. This will lead to correct tracking results, especially when the peak integration areas have originally been defined manually. 70 H _2_002

71 General Dynamics Method Center If you use drag & drop of spectra to the method tree as described in section Drag & Drop of Spectra to the Method Tree [} 26], spectra are loaded and displayed and time values are derived from time stamps. You may then continue with manual peak analysis right away or change the settings under Data as needed Analysis The analysis of kinetics data differs from other methods as individual signals may show individual intensity behavior. For example, in the data shown above the anhydrite signal decays exponentially whereas the acetic acid signal increases exponentially. One procedure could have been that the user looks at all spectra and decides which functional behavior should be assumed for which signal. To avoid this duty the Dynamics Center applies a number of built-in functions related to zero, first and second order kinetics as well as intermediates evolution. By means of comparing possible fit results via the AIC criterium, it decides the optimum solution. For a reference see Akaike Hirotugu, IEEE Transactions on Automatic Control 19 (6): , doi: /tac , MR Figure 6.34: The Analysis Dialog Window of the Kinetics Method. Currently the following functions are tested: (1) Linear decay (zero order reaction) H _2_002 71

72 General Dynamics Method Center (2) Exponential decay (first order reaction) (3) Exponential decay with a constant offset (4) Exponential build-up (first order reaction) (5) Exponential build-up starting from a constant offset (6) Intermediate (e.g. from A -k1->b -k2-> C) (7) Intermediate approaching constant offset 72 H _2_002

73 General Dynamics Method Center (8) Intermediate like (6) but with k1 >> k2 (9) Intermediate like (7) but with k1 >> k2 (10) Second order reaction (11) Second order approaching constant offset The first order fit functions (2-5) are pre-selected, the others will only be taken into account if the corresponding tic boxes are activated. The user also needs to select the error calculation method. The method Error estimation by weighted fit is strongly recommended. H _2_002 73

74 General Dynamics Method Center There is also an additional tab named Normalization. It allows you to rescale the peak intensities/integrals prior to curve fitting. The peak (or its integration area) used for normalization and the way to apply it need to be selected. A peak can either be taken from a given spectrum, set to a given value and then all spectra are normalized with the same factor. Or, peak selection and normalization is done on a per-spectrum basis. Figure 6.35: The Normalization Tab of the Analysis Dialog Window of the Kinetics Method. The combined use of normalization and divide by number of nuclei allows to represent the graphical results as concentration vs time and equivalent vs time. You can define the number of nuclei associated to a particular peak or integral by selecting the option Annotate from the peak popup menu. To obtain the menu, right click near a peak or peak integration area. The peak name and number of nuclei entered will become visible on the display and can be changed at any time. 74 H _2_002

75 General Dynamics Method Center View This component is used to specify details of the fit curve display and is mostly identical to the other methods. One difference is a new tab named Units & Label. Figure 6.36: Customizing the Fit Curve Display via the Units & Label Tab. Since all time values were originally given in seconds and a series of experiment might take a long time, the x-axis of the fit curve display may become unreadable with large numbers. A change of units, e.g. from seconds to minutes improves the display. An option to set a label for the Y axis (such as concentration (M)) is also offered in the Units & Label tab Report, Export These two components are identical to the other methods and serve to create PDF reports and exports in.xls, xlsx or.txt format Using 2D data With the availability of multiple receivers and fast 2D acquisition techniques like ASAP HSQC (e.g. Kupce E. and Freeman R., Magn. Reson. Chem., 2007, 45, 2-4), combined with non uniform sampling, it is possible to get 2D spectra in short times and use them for reaction monitoring. All descriptions given above (e.g. how to define peak integration areas and peak tracking) hold true for 2D data. The 2D peak tracking needs more computer resources, as the 2D data are needed in memory for the matching techniques. If the total number of 2D spectra is smaller than 64, the Dynamics Center loads all the spectra at once to save disk access time. In other cases repeated loading from the disk is performed. Different visualization tools are offered as appropriate. For example, instead of a 1D stacked plot it is possible to get a 3D display based on stacked 2D spectra. This 3D object can be rotated, scaled etc. and reaction profiles can be selected directly from it. H _2_002 75

76 General Dynamics Method Center 1 Figure 6.37: Example of using 2D Spectra for Reaction Monitoring 1 Move cursor to peaks to get reaction profiles selected. 6.5 The Cross Polarization Method Cross polarization from abundant spins (I) to dilute spins (S) is a double resonance technique (Kolodziejski W. & Klinowski J., Chem. Rev. 102, (2002)). Since CP is sensitive to inter-nuclear distances of I and S spins, as well as, molecular mobility, it is widely used to monitor dynamics in solids. The polarization transfer competes with T 1, T 1rho relaxation of I and S spins to the lattice. With some approximations, cross polarization as a function of contact time t is then governed by the equation. T IS is related to the speed of the polarization transfer and depends on distance and motion of I and S spins, while T1 rho of the I spins relates to slow lattice motions. Since T1 rho can be determined independently a second variation of the above equation where T1 rho is fixed, is offered Sample The Sample component is identical for all methods in the general dynamics method center and serves for input of information that is used for reporting purposes. 76 H _2_002

77 General Dynamics Method Center Data The Data component is similar for all methods within the general dynamics area. Possible formats that can be used are pseudo 2D or series of 1D spectra Analysis Analysis allows curve fitting of the data to the function given or other functions if the Python option is used. If the approximations are valid the fit usually works quite well. Figure 6.38: Successful CP Curve Fit of Alanine Resonances View, Report, Export These components are used to set the details of the result viewing, reporting and exporting and are more or less identical to the descriptions given in the previous sections. 6.6 The REDOR Method REDOR (Rotational Echo Double Resonance) is used to quantify the heteronuclear dipolar coupling between nuclei I (observe nucleus) and nuclei S (de-phasing nucleus) under MAS conditions. The dipolar interaction depends on the number and distances of neighboring atoms and involves the second moment defined as: In the case of isolated two-spin systems the second moment and dipolar coupling constant are related as: H _2_002 77

78 General Dynamics Method Center Sample The Sample component is identical for all methods in the general dynamics method center and serves for input of information that is used for reporting purposes Data The Data component is similar for all methods within the general dynamics area. Using the Bruker cpredori pulse program, pseudo 2D spectra are generated. They contain pairs of traces measured in interleaved mode at different dipolar de-phasing times. One trace results from an echo applied to the S spins and is called S0 signal. The other results from an echo applied to the S spins with a simultaneous de-phasing by the I spins. This signal is called S*. The dependence of the ratio (S0 - S*)/S0 as a function of the rotor speed can be related to the second moment M 2 as given above. Figure 6.39: REDOR Data Dialog Window. Only pseudo 2D spectra can be selected and the acquisition order, i.e. whether the pairs of traces are S0, S* or S*, S Analysis The functional dependence of (S0 - S*)/S0 on the rotor speed can be approximated by if only the initial data points are taken into account: or or The first two alternatives yield the second moment, whereas the third alternative yields the dipolar coupling constant (in the case of isolated two-spin systems). 78 H _2_002

79 General Dynamics Method Center Figure 6.40: REDOR Analysis Dialog Window. By setting an upper ratio of (S0-S*)/S0 you can control how many points are taken into account. It is assumed that the initial points are monotonously increasing in intensity otherwise an upper ratio may not be valid. H _2_002 79

80 General Dynamics Method Center View The View component is similar to other methods. From the fit curve display one can clearly see how well the approximation worked for the initial points. Figure 6.41: Typical REDOR Result. The Initial Points are Fit to the Second Moment. As described in section View [} 50] the popup menu of the fit curve display contains an entry named add external data. This can be used to load x, y value pairs from a comma separated (csv) file. REDOR curves simulated with the SIMPSON software can be displayed this way. Figure 6.42: Simulated REDOR Curves loaded from a csv File Report The Report component is similar to Report for any other method and is self-explanatory. 80 H _2_002

81 General Dynamics Method Center Export The Export component is similar to Export for any other method. It contains an additional tab to generate an output file that can be used as input for the Simpson (A General Simulation Program for Solid-State NMR Spectroscopy) software (Bak M., J.T. Rasmussen & N. C. Nielsen, JMR, Vol 147, 2, , (2000)). Figure 6.43: Export Dialog Window with an Interface to Simpson. The output file for Simpson is in textual form, pre-filled with available information and presented for further editing. H _2_002 81

82 General Dynamics Method Center 6.7 The Arrhenius Method In its current implementation the Arrhenius method serves to determine the activation energy from a series of diffusion experiments measured at different temperatures. In this sense the data analysis consists of two steps, in a first step in each experiment the diffusion constants are calculated as described in chapter The Diffusion Method. In a second step all calculated diffusion constants are used to fit the activation energy Sample The Sample component can be used to specify some simple information about the given sample. It is only used for documentation purposes and may be contained in reports or exports Data The Arrhenius only uses series of pseudo2d spectra. In the Data dialog the user can provide the spectra names and associated temperature values given in Kelvin. Figure 6.44: The Arrhenius Data dialog 82 H _2_002

83 General Dynamics Method Center The other tabs (Peaks, Integrals, Lists, Diffusion Par) are identical to the ones used in the regular Diffusion method. The first pseudo2d spectrum is loaded and peak picked accordingly, the first trace of this spectrum is displayed together with a slider that allows you to navigate to other traces of the first spectrum. See also 2 Drag & Drop of Spectra to the Method Tree [} 26] 2 Data [} 40] Analysis The analysis of each pseudo2d spectrum is identical to the analysis of a single diffusion spectrum as described in chapter: The Diffusion Method. For each of the peaks in the first pseudo2d spectrum diffusion constants are calculated based on the selected diffusion fit function. Figure 6.45: The Arrhenius Analysis Dialog with a tab named Arrhenius Analysis Once the first spectrum is analyzed its peaks are projected to the other pseudo2d spectra, local peak snapping is automatically applied and the calculation of the diffusion constants is repeated for all peaks in all the other spectra. The analysis dialog window contains an additional tab named Arrhenius Analysis. It has no other function than showing the Arrhenius formula which is used to fit the activation energy for each peak. The fit takes the diffusion constants as a function on temperature. Start parameters are calculated automatically. H _2_002 83

84 General Dynamics Method Center View Figure 6.46: Arrhenius Plot of one Peak This component is used to specify details of the fit curve display and is mostly identical to View in the other methods. When moving the mouse pointer to a peak (related to a certain trace in the first spectrum), the diffusion curves from all available pseudo spectra are automatically cumulated in a window and the Arrhenius curve derived from the individual diffusion constants is shown in a second window Report, Export These two components are similar to Report, Export in other methods and serve to create PDF reports and exports in.xls, xlsx or.txt format. Currently, the report also includes the individual diffusion fit results of each peak of each of the pseudo2d spectra while the export only includes data and results of the Arrhenius fit. 6.8 Recommended Pulse Programs FLEXlm name Spectrum List Function t1ir pseudo-2d vdlist inversion-recovery t1irpg pseudo-2d vdlist inversion-recovery cphirt1 pseudo-2d vdlist inversion-recovery cphsatrect1 pseudo-2d vdlist saturation-recovery 84 H _2_002

85 General Dynamics Method Center cpht1.av pseudo-2d vdlist saturation-recovery cpxt1 pseudo-2d vdlist inversion-recovery cpxt1.av pseudo-2d vdlist inversion-recovery satrecechot1 pseudo-2d vdlist saturation-recovery satrect1 pseudo-2d vdlist saturation-recovery satrect1.av pseudo-2d vdlist saturation-recovery satrecechot1.av pseudo-2d vdlist saturation-recovery cpmg pseudo-2d vclist decaying-exponential cpht1rho pseudo-2d vplist decaying-exponential cpht1rho.av pseudo-2d vplist decaying-exponential cpxt1rho pseudo-2d vplist decaying-exponential cpxt1rho.av pseudo-2d vplist decaying-exponential zght1rho pseudo-2d vplist decaying-exponential H _2_002 85

86 General Dynamics Method Center 86 H _2_002

87 Protein Dynamics: Basic Relaxation Analysis 7 Protein Dynamics: Basic Relaxation Analysis 7.1 Introduction Each relaxation method shown on the method tree contains the same components Sample, Data, Analysis, View, Report, Export. These components have to be executed in this order. After a successful execution the corresponding nodes are green color-coded on the method tree. A red color indicates that a component is currently executing, e.g. doing a longer calculation. Any subsequent components can only be executed if the previous component is shown in green. In the example below state A shows the Diffusion method just opened. State B indicates that Sample and Data have been successfully executed while Analysis is currently in progress. The next following component (View) can only be executed after Analysis has turned green. It is however possible to execute components of other methods in parallel. Figure 7.1: Color Codes Indicate the State of a Method. H _2_002 87

88 Protein Dynamics: Basic Relaxation Analysis 7.2 Sample The Sample component is used to provide information about the sample used. Figure 7.2: Tab Oriented Dialog Window to Describe the Sample. The information to be provided is mostly used for report purposes. When moving the cursor into a field a screen-tip text provides some additional information. Some of the given sample information can be used during the analysis or for display purposes. In particular this holds for the AA sequence tab under which the format (FASTA, SEQ) and name of an amino acid sequence file can be specified. All relaxation methods offer a histogram display of results (T 1, T 2, etc.) versus sequence. Since the results relate to individual peaks the peak annotations must contain residue information otherwise proper histogram displays are not possible (also see section View below). For example, proper peak annotation would be ALA 10 or ALA/10 or ALA [10]. The Structure tab is used to specify the name of a pdb file. The structure is needed if anisotropic modelling of dynamic parameters is to be done or modelling results shall be shown on the structure (see next chapter). 88 H _2_002

89 Protein Dynamics: Basic Relaxation Analysis 7.3 Data The Data component is used to define details related to the spectra. Figure 7.3: Data Dialog for the Selection of Spectra. Bruker has released pulse programs for relaxation methods, the once used here generate pseudo 3D spectra with the relaxation dimension in F1. Mixing times (or field strengths) are then automatically retrieved from spectra parameters (further details below). However, if you provide a series of 2D spectra names you must also provide the mixing times (or field strengths) in the dialog window. The order of the planes is as stored in the pseudo 3D, the order of the 2D spectra as specified in the Data dialog window. In both cases planes or spectra do not need to be sorted, e.g. according to mixing times. All relaxation analysis methods are based on peak intensities/integrals. Peak picking and peak integral calculation therefore play a central role. The Peaks tab in the Data dialog window offers several choices. Peaks can be automatically picked or existing XML peak lists can be used. The automated picking has a built-in adaptive peak picking threshold estimation. Noise levels are determined for each data point. Simple descriptive statistics is used to check if columns or rows in the spectrum contain an unusual number of peaks. If a valid sequence file was specified the overall number of peaks can be well estimated. However, the most important source of information is however that the spectra are available N times (N planes of a pseudo 3D or N 2D spectra). The signal-to-noise in these spectra differs according to experimental parameters (e.g. mixing time) but a peak should occur in the majority of spectra. Note: To optimize a peak list manually and use it for relaxation analysis, use the options in the peak popup menu and save to disk. Then chose the option use peak list at spectrum under Data/Peaks. Using a peak list from another spectrum makes sense if there is a well picked reference spectrum whose peaks can be used in other spectra, for example a good 2D HSQC peak list can be used in all planes of a pseudo 3D spectrum. If such peak lists are imported, it is advised to allow so-called peak snapping. This means that imported peaks are moved to near experimental local maxima when possible. One snapping algorithm combines a global shift and local nearest neighbor search. This works well if the global shift can be automatically detected looking at some outlying peaks. In some cases the more simple snap using a local neighbor search gives better results provided the snap radius does not need to H _2_002 89

90 Protein Dynamics: Basic Relaxation Analysis be selected so large that snapping to neighboring peaks becomes a problem. In cases where peak positions are very stable along different spectra (planes) it is possible to do snapping in the first spectrum, then just copy peaks to the others. Identical digital resolution and calibration of all spectra is required. For relaxation parameter determination peak intensities (preferred option) or peak integrals calculated by region integration (regions pre-defined but can be changed interactively), shape integration (shapes down to half height determined automatically) or peak de-convolution can be used. In most cases, especially if peaks are isolated, there is no big influence on the fitted parameters. The peak de-convolution method uses preselected peaks. Therefore a careful (manually optimized) peak selection is advised. Different shapes Gaussian, Lorentzian, mixed) can be applied, but suitable start parameters for the peak line widths should be given. Figure 7.4: Comparison of a T1 Fit Based on Intensities (A), Area Integrals (B), Shape Integrals (C) and De-convolution (D), Differences of T1 are Less Than 1%. Note: Currently, TopSpin does not store peak integrals in peaklist.xml. Therefore, peak integrals are re-calculated each time a project is opened. The spectra popup menu contains measure distance as a graphical tool to estimate it. The used de-convolution algorithm assumes no correlation of parameters of different peaks which speeds up the calculation. Nevertheless, minutes are typically needed for a full pseudo 3D spectrum with 12 planes and about 100 peaks each with Gaussian shapes which is recommended as default. The variable Gaussian/Lorentzian shape needs 2-3 times longer. You must wait until this is finished before continuing with Analysis of the current method. You may however work with other methods in parallel. Illegal operations, e.g. closing the 90 H _2_002

91 Protein Dynamics: Basic Relaxation Analysis spectrum while the de-convolution, is calculating are blocked during this time. It is possible to cancel the peak de-convolution but the current plane of a pseudo 3D spectrum or current 2D spectrum of a series of spectra is finished before the de-convolution is really stopped. If assignments are available from elsewhere they can be imported. Currently BMRB (.str) or XEASY (.peaks) formats are supported. BMRB files may have 8 or 9 columns in the assignment section (lines following the _Atom_shift_assign_ID to _Chem_shift_ambiguity_code lines). XEASY files may contain peak annotations in each following line of a peak, for a description see for example: Only a simplified automated peak snapping is applied to find the closest peaks if assignments are imported. The Lists tab relates to handling the mixing times (or field strength) and is strongly context sensitive. In case of T 1, T 2,T 1rho and R ex data the mixing times (or field strengths) are either taken from vdlists (if Bruker pulse programs have been used) or the user has entered the values by hand under the Spectra tab. When acquiring T 2 data with Bruker pulse programs, there is no vdlist but a vclist which first must be converted into a vdlist. This conversion is done automatically and uses the number of mixing times, loop duration and constant duration which are read from the acquisition parameters and pulse program dependant. For example, when using hsqct2etf3gpsi3d for example the loop duration is equal to D31 but when using trt2etf3gpsi3d or cpmg it is calculated from P2 and D20. In case of Rex spectra peak intensities/integrals need to be converted into decay rates first (Mulder et al., J. Am.Chem. Soc., Vol 123, (2001)). This conversion needs the total mixing time (length of CPMG pulse train) and is read from the acquisition parameters if Bruker pulse programs have been used, otherwise it must be entered here. The question remaining is how to treat repetition experiments. It is recommended to measure the one or other mixing time more than once to check for reproducibility of the data. Repeated experiments can be kept or collapsed into mean data. Repetition experiments are also evaluated when calculating uncertainties of peak intensities or peak integrals (see next section). Currently, 2 different options are offered to estimate systematic errors. The estimate can be based on variance averaging using peak intensities/integrals. All repetition experiments and peaks are taken into account, and the square root of the average variance is then assigned to all the peaks at each mixing time. The other alternative is to use all repetition experiments to calculate the largest difference of peak intensities/integrals per peak and assign this as a systematic error per peak for each mixing times. Variance averaging balances over- and underestimates of systematic errors whereas the difference method slightly overestimates systematic errors. This may have an impact on the model free modeling. H _2_002 91

92 Protein Dynamics: Basic Relaxation Analysis Figure 7.5: Data Display with Slider. Finally, the specified data are loaded and displayed. In case of pseudo 3D spectra the first plane, and in case of a series of 2D spectra the first spectrum are shown. A data-slider which allows you to select other planes/spectra appears. The slider can be regarded as a player which contains buttons forward/backward play, fast forward, fast backward and stop. Another button offers additional settings to be selected. When several methods are active with several data displays open then several sliders will be shown, one slider for each method. These sliders normally work independently of each other but can also be correlated, see options offered via the settings button. 92 H _2_002

93 Protein Dynamics: Basic Relaxation Analysis 7.4 Analysis The Analysis component is used to fit the peak integrals to particular functions and extract the relevant relaxation parameters. It can be executed if the Data component was already successfully executed before. Figure 7.6: Example of the Analysis Dialog Window, here T2 Relaxation. The fit function is indicated in the dialog window. In case of the T 1rho and T 2 methods a single pre-defined function is available. Start values for the parameters to fit must be provided. In case of T 1 there are different alternative functions. Start values for the parameters to fit are calculated automatically. You can also define your own function in Python style. Then any function is given in a defined notation. The function value must be called y, the variable is x and fit parameters are given as an array called p. There may be constants involved if entered as numbers. y = f (p, x, constants) An example would be y = p [0] exp (p[1] x + p[1] which would correspond to y = I exp(t - x) + C where I, T and C are fitted. Comma separated start parameter values must be given, e.g. 1.0e7, 5.5, 0 in this example. The amplitude, here p[0], must be provided in numbers as found for intensities in the spectrum, e.g. 1.0e7. While moving the cursor in a 1D or pseudo 2D spectra the intensity is shown. Properties of peaks obtained via the peak popup menu show peak intensities. In 2D spectra the contour level suggests a typical intensity. It is usually sufficient to estimate the right order of magnitude of the amplitude. Later on, when displaying histograms or reporting fit results (see View, Report below), p[1] is regarded as the most relevant parameter. As fit parameter unit any comma separated strings can be provided, if there is no unit, specify none, e.g. none, s, none in the above example. H _2_002 93

94 Protein Dynamics: Basic Relaxation Analysis The following example shows an analysis of J-modulated 1H-15N HSQC spectra. Figure 7.7: Example of the User Defined Function to Analyze J-modulated 1H-15N HSQC Spectra. 94 H _2_002

95 Protein Dynamics: Basic Relaxation Analysis If applied to imported and snapped peaks to a pseudo 3D spectrum the following result is obtained. Figure 7.8: Results of fitting a User Defined Function to J-modulated 1H-15N HSQC Peaks. Three options related to error estimation are offered. Use the Y data (peak intensities, integrals) under the assumption that their uncertainties are unknown but equal for all Y values. The non-linear fit determines errors for the fitted parameters from the inverse of the un-weighted curvature matrix (second derivatives of chi-squared). The final chi-squared itself is arbitrary. The second option applies if individual uncertainties of the Y values are known and are passed to the non-linear fit. The errors of the fitted parameters are then calculated from the inverse of the weighted curvature matrix. The uncertainties themselves are derived from the standard deviation of the noise in each plane/trace/spectrum and, if available, differences of Y values in repetition experiments. Use a Monte Carlo simulation. The non-linear fit is performed 500 times with the input Y data varied according to a normal distribution, with a standard deviation equal to the uncertainties of the Y values. In order to provide fitted parameters within a confidence interval a confidence level needs to be given. The confidence interval of a fitted parameter is then calculated by multiplying the error of the fitted parameter with a factor taken from the inverse of Student-s-T cumulative distribution at given confidence level and number of degrees of freedom. The confidence interval calculates as fitted parameter +/- fitted parameter error. The larger the chosen confidence value, the larger is the interval. H _2_002 95

96 Protein Dynamics: Basic Relaxation Analysis 7.5 View The view component allows viewing of the obtained results is a central component of any method. Important is to correlate spectra/peaks with parameter fits shown as individual fit curves or together with the amino acid sequence. The user may select which types of displays to view, the correlation between all display windows is provided automatically. Figure 7.9: View Dialog Window to Customize the Result Display. Interactive fit displays are generated according to cursor movement either on spectrum or histogram. The fit curve of the peak close to the cursor is displayed in an internal or external window. Depending on the selected option it is enough to just move the cursor close to a peak or an additional left mouse button click is needed. Error bars (uncertainties of peak integrals, see previous section) can be drawn on the fit curves and the exponential decay curves can be shown in logarithmic form. The fit curve display is a regular display object and can be scaled, zoomed etc. Fitted relaxation parameters, as related to an amino acid sequence, can be displayed in histogram style and may include error bars that correspond to the errors of the fitted parameters, e.g. T 1. The histogram is a regular display object and can be scaled, zoomed etc. Finally, chi-squared of each fit divided by the sum of all Y values taking part in that fit can be displayed as a histogram. It serves as a visual diagnostic tool to estimate the quality of input data and fit. A typical display with spectra/peaks, individual fit curves and histogram may be as follows, whereas the green point indicates current plane/spectrum: 96 H _2_002

97 Protein Dynamics: Basic Relaxation Analysis Figure 7.10: Typical Result Display with Spectrum, Histogram and Fitted Curve. The cursor can be moved in the spectra display or on the histogram. Corresponding fit curve are displayed according to the selections made in the View dialog. If multiple projects are open and multiple spectra/histogram displays are on screen they are all correlated. The X variables (e.g. mixing times in the vdlist) may be in arbitrary order and correspond to the order of planes/spectra. The fitting procedures need peak integrals in sorted order which is handled automatically. To see which plane/spectrum relates to which point in the fit curves, a green marker is shown. This marker jumps from point to point if you switch to other planes/ spectra using the spectra slider. H _2_002 97

98 Protein Dynamics: Basic Relaxation Analysis Fit display objects As mentioned the fit display objects are regular display objects and can be scaled, zoomed etc. The fit displays also contain context sensitive popup menus. Figure 7.11: Fit Display Objects also have Context Sensitive Popup Menus. Context sensitive popup menus are available after a right mouse button click. If the cursor was near a point in the fit curve (it will be highlighted with a red color) the popup menu offers: Delete point The current point is deleted and the fit curve is recalculated. Undo latest delete The point deleted last is restored and the fit curve is recalculated. This can be repeated until all previously deleted points are restored. Delete point apply to all A point in a fit curve is often deleted if the data of that particular mixing time are bad. In this case it makes sense to delete this point in all fit curves. Note, that there is no global undo that would restore a point in all curves. Delete point in range All points in a range defined with the cursor (left click, drag, release) are deleted. With undo latest delete you can restore deleted points one by one. Delete point in range, apply to all All points in a range defined with the cursor (left click, drag, release) are deleted in all fit curves of all peaks. 98 H _2_002

99 Protein Dynamics: Basic Relaxation Analysis Remarks: Manipulation of fit curves is only intended in special cases, for example, if one wants to see if a result can be obtained after deleting bad points. Manipulations of the fit curves are indicated among the properties of the fit curve, but are lost when the program closes. The only justified case would be to delete all points in all fit curves at high mixing times if the signal has already decayed into the noise. But this can better be achieved by lowering the TD value in the Data section. When the right mouse button is clicked outside any fit point (no red highlight visible) a popup menu with the following options is offered: Toggle This is the standard display toggle used to show the fit curve in full screen. Visibility Controls the visibility of individual display objects. This is only used occasionally since there are several other ways to customize the display (Toggle, View, Suspend on the method tree). Undo latest delete point The point deleted latest gets restored and the fit curve is recalculated. This can be repeated until all previously deleted points are restored. Properties Some properties of the current fit curve are shown. These include the used X and Y values, fit results, goodness of fit and some peak information. Close Closes the fit display, however it will reappear as soon as you move the cursor to another peak in the spectrum or another histogram item if visible on screen. To permanently switch off the fit curve display of a method re-execute the View component and chose the proper setting there. Histogram display objects The histogram display compares, for example, relaxation parameters with the amino acid sequence. This is possible if the peak to which the parameters belong contain residue numbers as part of their annotation, e.g. ALA 10 or ALA [10] or even just 10. An amino acid sequence file must have been specified under Sample of a method. If no sequence has been specified, or if it cannot be loaded properly, a pseudo sequence 1...n is assumed and displayed. The correlation between histogram, spectrum and fit curve display still works. However, if peak annotations do not contain any number then they cannot contribute to the histogram. They also cannot contribute to the histogram if a sequence is properly loaded (and thus defines how many items the histogram has), but the peak annotation contains a higher number. H _2_002 99

100 Protein Dynamics: Basic Relaxation Analysis 7.6 Report A PDF report that contains loaded information, as well as results, can be generated in PDF format. Graphical components include the current spectrum and fit curves display. Numerical components include the sample, peak and fit information. The user may select desired components in a dialog window. The name of the pdf file must also be specified. At the end of the report generation AcroRead is automatically launched to display the report. Some versions of AcroRead do not display the PDF file if it was not specified with its absolute path name, e.g. c:\tmp\test.pdf. If you don t have AcroRead on your computer an alternative PDF display program may selected under Config/Preferences/Default PDF Viewer. 7.7 Export Information, especially peak information and fit results, can be exported to a file on disk. The supported formats are text and xls (or xlsx). This allows users to present data with standard tools, e.g. chart diagrams in EXCEL or to load peak integrals into other software packages to repeat the fit calculations, or fit other functions not offered in the Dynamics Center. Note that Export differs from Report in several details. Numerical values are shown with more digits, graphical objects are not exported. The numerical section may contain extra information, e.g. the errorscale values that have been multiplied to the errors of the fitted parameters to get confidence bounds. 7.8 Further Information Recommended Pulse Programs Bruker has released a number of pulse programs to acquire relaxation data. Examples are (all resulting in pseudo 3D spectra) 15N / Channel f3 HSQC version TROSY version NOE hsqcnoef3gpsi trnoef3gpsi T1 hsqct1etf3gpsi3d trt1etf3gpsi3d T2 hsqct2etf3gpsi3d trt2etf3gpsi3d T1rho hsqctretf3gpsi3d trtretf3gpsi3d Rex hsqcrexetf3gpsi3d trrexetf3gpsi3d Summary of Fit Functions The following shows the fit functions used. T 2 I(t) is a decay curve of y-values (peak intensities/integrals) t is the x-variable (mixing time) I0 (amplitude at t=0) and T2 (decay constant) are fitted start parameter for I0 is the y-value at lowest mixing time (automatically chosen)start parameter for T2 can be given by the user (e.g. 0.2s) 100 H _2_002

101 Protein Dynamics: Basic Relaxation Analysis T 1 Exponential decay: Saturation recovery: Inversion recovery: Partial inversion recovery: I(t) is the decay or build up of y-values (peak intensities/integrals). t is the x-variable (mixing time). I 0 (amplitude at t=0) and T 1 are fitted. start parameter for I 0 is the y-value at lowest mixing time (automatically selected). start parameter for T 1 can be given by the user (e.g. 0.5s). If the inversion is not fully performed for all signals in an inversion recovery experiment it is advised to select the partial inversion recovery function. For reasons of completeness the exponential decay function is also offered with a constant offset term that is also fitted. This function can also be used for other purposes, e.g. proton exchange. T 1rho Exponential decay: I(t) is a decay curve of y-values (peak intensities/integrals). t is the x-variable (mixing time). I 0 (amplitude at t=0) and T 1rho (decay constant) are fitted. start parameter for I 0 is the y-value at lowest mixing time (automatically selected). start parameter for T 1 can be given by the user (e.g. 0.15s). As part of the T 1rho analysis T 2 values are also approximately calculated (Palmer at al., Chem. Rev. 106, (2006)). with H _2_

102 Protein Dynamics: Basic Relaxation Analysis B 1 is the RF field strength, the resonance offset. R ex (slow exchange limit - Tollinger M. et al., J. Am. Chem. Soc., 123, (2001)) R( cp ) is a curve of y-values, in this case decay rates obtained from peak intensities/integrals cp is the x-variable, i.e. derived from RF field strengthr20 (amplitude at cp =0), ex (frequency difference of exchanging sites) and ex (exchange rate) are fitted. Start parameters for all must be given by the user. R ex (fast exchange limit) (Wang C. et al., J Biomol NMR, 21(4), (2001)). R( cp ) is a curve of y-values, in this case decay rates obtained from peak intensities/integrals cp is the x-variable, i.e. derived from RF field strength R 20 (amplitude at cp =0), 2 ex - p a - p b (frequency difference of exchanging sites, p a p b populations) and ex (exchange rate) are fitted. Start parameters for all must be given by the user. The decay rates are obtained from peak intensities/integrals (Mulder A. et al., J. Am.Chem. Soc., Vol 123, (2001)). T/2 is the total mixing time. 102 H _2_002

103 Protein Dynamics: Modeling Backbone Dynamics 8 Protein Dynamics: Modeling Backbone Dynamics 8.1 Introduction The basic relaxation methods include NOE, T 1, T 2, T 1rho and R ex experiments. Following the strategy most often described in the literature (Fushmann D., BioNMR in Drug research, 283 ff, (2002)) a general analysis of the protein backbone dynamics requires the combination of at least the NOE, T 1 and T 2 methods. This combined method is named NOE/T1/T2 on the method tree. Figure 8.1: The Combined NOE/T1/T2 Method is Part of the Method Tree. The difference in this method compared to the basic methods, is that this combined method uses the data already contained in the NOE, T1 and T2 methods. Therefore, the sample information is therefore already known and NOE/T1/T2 does not contain Sample on the method tree. 8.2 How to Execute the NOE/T1/T2 Method The NOE/T1/T2 method uses data and results from the NOE, T1 and T2 methods. Therefore, these three methods must have been set-up and executed at least up to Analysis and then saved to project files (right mouse click to NOE, T1 and T2 and Save from the popup menu). If data are available at multiple field strengths, a set of NOE, T1 and T2 projects must have been saved for each field. To run the NOE/T1/T2 method, execute NOE/T1/T2/Data and select the number of field strengths and the names of the project files (which you saved before) to be used. H _2_

104 Protein Dynamics: Modeling Backbone Dynamics Figure 8.2: Data Dialog Window to Select Project Files to be Used for NOE/T1/T2 Modelling. It doesn t matter at this point whether the data from these projects is in memory. Alternative 1: If you start with a blank screen an automated loading takes place after the Data dialog window is closed with OK. One spectrum after the other is loaded and analyzed. Normally, the Dynamics Center performs long lasting operations (peak de-convolution or curve fitting with Monte Carlo error estimation) asynchronously. This means that a progress bar appears but the program stays active. While the progress bar moves on you can do interactive work or continue with other projects. If the projects are loaded via the NOE/T1/T2 method this asynchronous behavior causes problems, thus it is switched off. While a progress bar is visible you have to wait until an operation is finished. However, the loading works fine if the method tree shows green leafs. 104 H _2_002

105 Protein Dynamics: Modeling Backbone Dynamics Figure 8.3: Successful Loading of a Group of NOE, T1 and T2 Projects. This alternative is especially recommended when no long lasting operations (peak deconvolution, Monte Carlo calculations) are involved. Alternative 2: First load the projects one-by-one (right click to NOE, Open, then right click to T1, Open,...) and execute each method up to Analysis. This way you can better control which actions are done. If NOE/T1/T2 is executed afterwards, you will be asked if you want to use the NOE, T1 or T2 data already in memory, otherwise the projects are loaded again in automation like in alternative 1. If data is various fields are available use the Add to tree option from the popup menu to duplicate the NOE, T1 and T2 methods on the method tree. After the specified projects are loaded, some internal checks are performed. All spectra should refer to the sample and the amino acid sequence. It does not make any sense that different amino acid sequences had been defined for the NOE, T1 and T2 methods. Preferably, all participating spectra contain the same number of peaks annotated the same way. However, this could be too restricting. Therefore the peak list of the first T1 spectrum is regarded as most relevant. All peaks on the peak list are used if the corresponding peaks are also found in the other spectra. If different numbers of peaks are detected a corresponding message is presented. You may then continue with NOE/T1/T2/Analysis to do the various modelling calculations. Even though the NOE, T1 and T2 data may have been analyzed before, it is still allowed at any time to apply changes to the NOE, T1 and T2 data, e.g. change a peak. NOE/T1/T2/ Analysis must then be re-executed to synchronize results. The View, Report and Export components work as usual, more details are given below. As with other methods the NOE/T1/T2 method shows a popup menu after right clicking on it. It also contains the regular entries Open, Save, Save As, Close, Suspend and Resume. In contrary to other methods an Add to tree entry is not active in this popup. The reason is simply to restrict complexity. The Save or Save As options should be used to save the current NOE/T1/T2 method into a project file on disk. As usual it contains information and settings of the method but not the modelling results itself. H _2_

106 Protein Dynamics: Modeling Backbone Dynamics 8.3 Extracting Dynamic Information from T1, T2 and NOE R1 (=1/T 1 ), R2 (=1/T 2 ) and NOE are related to combinations of a spectral density functions J at 5 different frequencies (0), J(ω N ), J(ω H ). J(ω H + ω N ), J(ω H - ω N) of X-H bond motions. Since 5 unknown spectral densities cannot be determined from 3 experimental data points, so-called three reduced spectral densities J(0), J(ω N ), J(0.87ω H ) were introduced which can directly be calculated from R 1, R 2 and NOE. With certain assumptions the difference between measured R 2 values and R 2 values predicted via the reduced spectral density functions can be explained as chemical exchange contributions to R 2. Unfortunately, the interpretation of the reduced spectral densities in terms of global and local motions is not straightforward, but even so a semi-manual method called Lipari-Szabo mapping exists. Most widely used in turn is the model-free formalism introduced by Lipari and Szabo which separates the global tumbling motion (described by a global correlation time) from a local motion of a X-H bond, characterized by an order parameter S2 (ranging from 0 to 1, 1 meaning no local motion) and a local correlation time. The model-free formalism was later extended to allow two different local motions on two different time scales per X-H bond. Furthermore the model-free models depend on whether the global tumbling of the protein can be described by one isotropic diffusion constant or rather by an axially symmetric or even anisotropic diffusion tensor. In the later cases the components of the diffusion tensor must first be determined. A possible approach is to evaluate the ratio between J(w N )/J(0) which does not depend on local motions, or R 2 /R 1. The protein structure is needed in these cases. If the protein structure is not available it is still possible to estimate the ratio of parallel and perpendicular components of the diffusion tensor. If this ratio is small, say < 1.2, then assuming global isotropic tumbling is justified. A substantial number of computer programs doing model-free analysis are available. It is a common practise to offer sets of model free models and fit each model to each residue. The models have different numbers of parameters, usually up to 3 are determined by the fit, others are held constant. The model that fits best to a certain residue is regarded to be 2 relevant for that residue. As a criterion not only but for example the AIC value, where k is the number of fitted parameters is evaluated. It should be mentioned at this point that different major modelling strategies exist. The more simple one assumes that it is possible to find residues that do not show local motions or relaxation exchange by inspecting T 1 /T 2 and NOE values. From these the global correlation is estimated and kept fixed. Another approach is to keep the global correlation time non-fixed and modify it after each modelling step. A larger number of iterations between modelling and global correlation time calculation is done in this case. 1 D. Fushman. BioNMR in Drug Research, p. 288 ff. 2 Farrow et al., J. Biomol NMR 6, , (1995). 3 Henkels et al., Biochemstry, 46, , (2007). 4 Andrec et al., J. Biomolecular NMR, 18, , (2000). 5 Lipari & Szabo, J. Am. Chem. Soc., 104, , (1982). 6 Clore et al., J. Am. Chem. Soc., 112, , (1990). 7 Walker et al., J. Mag. Res., 168, , (2004). 8 Tjandra et al., J. Am. Chem. Soc., 117, (1995). 9 D. Fushman. BioNMR in Drug Research, p. 296 ff. 10 Akaike, H., Proceedings of the 2nd International Symposium on Information Theory, , (1973). 106 H _2_002

107 Protein Dynamics: Modeling Backbone Dynamics 8.4 Strategy Used with the NOE/T1/T2 Method A typical and simple protocol for protein backbone relaxation analysis reads as: Calculate the three reduced spectral densities J(0), J(w N ), J(0.87w H ). Use them to estimate local correlation times, order parameters and relaxation exchange constants for each residue. Estimate the global isotropic correlation time c and a subset of residues that do not show relaxation exchange. of the diffusion tensor using Check NOE values to exclude residues that undergo fast local motions, e.g. those with NOE > 0.7. Check low T2 values to exclude residues that show conformation exchange. Since low T2 values can also be caused by anisotropic tumbling one may rather check combined T1 and T2 values (e.g. from Tjandra/tjandra_practical_relax.pdf). simultaneously with for example f=1.0, SD is the standard deviation of T2 and Check whether the isotropic diffusion model can be assumed. with for example n=3.0. Case A: Isotropic diffusion can be assumed. Fit individual models of motions to individual residues using T 1, T 2 and NOE values per residue. The estimated overall isotropic correlation time is held fixed. The models are called M1 - M5. No protein structure is required. Case B: Anisotropic diffusion must be assumed. In this case the protein structure is needed since relaxation depends on the orientation of the NH bond vectors relative to the axes of the diffusion tensor. The diffusion tensor of the protein is first determined by fitting J(w N )/J(0) or R 2 /R 1 for all residues. The single overall correlation time used in M1 - M5 now is replaced by five individual correlation times which are derived from the components of the diffusion tensor. These correlation times are then held fixed and are incorporated into the model free models which are called T 1 - T 5. These models also contain five coefficients that depend on the orientation of the NH bond vector relative to the diffusion tensor. These coefficients have to be determined for each residue individually. Remarks on the diffusion tensor: Various alternative procedures are described in the literature. Besides isotropic and anisotropic diffusion an intermediate axially symmetric diffusion is considered, for example. There are also different ways of determining the components of the diffusion tensor, e.g. in a combined modelling of local motions per residue and global modelling of the diffusion tensor) (Cole et al., J. Biomolecular NMR, 26, , (2003).). Another approach is to fit an individual isotropic diffusion model to each individual R 2 /R 1 pair (Lee et. al., J. Biomolecular NMR, 9, , (1997)). Remarks on the model fitting: There are also various ways of interpreting the model fitting. Most common is to regard a model as valid if it best fulfills some statistical criteria. It has been observed, even with simulated data, that various fitting algorithms yield different results. One reason is that a fit of a model function with two or three parameters to only three data points works less well, then for example, 12 data points were available. Another reason is that the parameters are H _2_

108 Protein Dynamics: Modeling Backbone Dynamics constraint and there are different ways to incorporate constraints into the fit algorithm. Fits should always incorporate uncertainties of the data points. The determination of the uncertainties can be done in different ways which influences the fit results. Finally, some models just do not fit. For example, if a residue shows conformational exchange, models that do not contain an exchange parameter do not fit. Running the fit algorithm without constraints would then usually yield results that are obliviously false, e.g. an order parameter > 1. But since constraints are typically applied some other combinations of fit parameters are obtained that look reasonable at first sight. The hope is that the statistical evaluation select them as inacceptable results. The Dynamics Center uses a combination of Simplex and Levenberg-Marquardt algorithms. Random selection of a larger number of start parameters, e.g. 1000, is possible and suggested. Constraints are defined but not used during the minimization. However, results that violate constraints are rejected. Thus, it happens that a model does not yield any reasonable result at all. This is a clear indication that the model is not applicable. Remarks on using multiple fields: From the preceding explanations it is obvious that having NOE, T1 and T2 available at different fields improves the modelling. Up to 5 different field strengths can currently be used with the NOE/T1/T2 method. Reduced spectral densities and estimated parameters thereof are calculated per field strength. The calculated global correlation time is averaged over different fields and then used for the further modelling. One should however check that data and curve fitting are of good quality in all fields. The display capabilities allow you to verify this. A typical problem that happens quite often is that a reference peak list is imported to data but due to differences in the spectra the import is insufficient in some cases. As a consequence the curve fitting is bad for some of the peaks, but these may influence the estimation of the diffusion tensor significantly. 8.5 Performing analysis with the NOE/T1/T2 method After having selected Analysis, a dialog window to setup the calculations is shown. Different categories are accessible under different tabs Settings Some global settings can be set under this tab. The XH bond length (usually NH) is 1.02 Angstrom by default. The N chemical shift anisotropy, CSA, is often taken as an average of -160 ppm. In reality it varies from residue to residue and can cover values from -120 to 215 ppm. The number of iterations refers to the random selection of start parameters when doing the actual modeling (see below). A value of 0 means that no extra selections are done, we recommend value of 500 or Override calculated errors with defaults needs some explanation. The source of all calculations are the T 1, T 2 and NOE spectra. The errors of the extracted peak intensities or integrals depend on signal-to-noise and repetition experiments. Repetition experiments are usually available for some of the mixing times of the T 1 and T 2 experiments, but not often for the NOE experiment. The T 1 and T 2 relaxation parameters errors are obtained via T 1 and T 2 fits, which take into account the intensity/integral errors. In contrast the NOE value is only derived from a ratio of two peak intensities/integral and the error of the NOE value results from an error propagation calculation. A consequence of this procedure is that the NOE errors are often much smaller than the T 1 or T 2 errors. Also, the absolute errors may become quite small. 108 H _2_002

109 Protein Dynamics: Modeling Backbone Dynamics Figure 8.4: The Settings Tab is Used to Define Some Global Parameters. When running the modelling it is often seen that the NOE values are less well reproduced than T 1 or T 2 values. With small NOE errors they then dominate the final c 2 values of the fits. In fact the final c 2 values may go up dramatically like up to 106. Other models that reproduce the NOE values (but not necessarily the T 1 and T 2 values) slightly better, come out with a much smaller final c 2 value and will always dominate in a model comparing procedure. An alternative provided here is to override the calculated errors with default errors which can be specified as percentage numbers separately for T 1, T 2 and NOE values. The errors are then obtained by taking each T 1, T 2 and NOE value and by multiplying it with the percentage number, i.e if 2% was selected TauC Calculation of the global isotropic correlation time c. Figure 8.5: The TauC Tab is Used to Select Proper Residues for the Global TauC Calculation. H _2_

110 Protein Dynamics: Modeling Backbone Dynamics Suitable residues can be selected according to 2b). The value c is then obtained by fitting the selected T 1 /T 2 ratios t0 (Kay et al., Biochemistry, Vol. 28, No.23, 1089, page 8974) The final c value presented is the average over all fitted values. Alternatively, c is estimated as an average over local c calculated for each residue as: Again the selected residues are selected. Other suggestions described in the literature (Fushman et. al., J. BioMol NMR, 4, 61-78, (1994).), e.g. taking all residues, but use R1 and R2 values corrected for high frequency components and/or exclude some local c values from the final averaging, are not used here. Usually, both calculations yield very similar results Reduced SD Calculate reduced spectral densities, estimate, Rex and S 2. Figure 8.6: The Reduced SD Tab is related to Spectral Density Calculations. The reduced spectral densities are always calculated since they are also used internally. Based on selected T1/T2 pairs they are also used to estimate the rotational anisotropy. A quantity r defined as in which R 1 and R 2 are taken from R 1 and R 2 corrected for high frequency motions (Fushman. BioNMR in Drug Research, p. 294.) is calculated for all selected residues. The maximum and minimum found values for r correspond to NH vectors pointing parallel or perpendicular to the rotational diffusion axis. and of the diffusion tensor can be calculated as 110 H _2_002

111 Protein Dynamics: Modeling Backbone Dynamics and If is small, say < 1.2, then the overall motions can be considered as isotropic. Isotropic modelling does not need any information about the molecular structure. The user can also select if R ex and S 2 should be directly estimated from the reduced spectral densities for each residue. If multiple field strengths are available reduced spectral densities, c, and are calculated per field strength and are, for example, shown in reports and when exported. The further isotropic modelling (see below) uses an averaged t c Isotropic Modelling Fitting NOE, R1 and R2 of each residue to one or more isotropic models. A number of model-free models have been proposed in the literature. Various software packages offer various numbers of these models for fitting. Following the majority we offer 5 most commonly used models for the spectral density functions. M1 S 2 fitted, global isotropic correlation time t c held fixed M2 S 2, e fitted, global isotropic correlation time t c held fixed, formula given here is the numerically stabilized form (d Auvergene, Protein Dynamics, p. 91.). M3, like M1 but a R ex term is added to the R 2 calculation. M4, like M2 but a R ex term is added to the R 2 calculation. M5 As an approximation the fast local motion is assumed to be very fast, i.e., f -> 0, the global isotropic correlation time c held fixed, fitted are S 2, S f 2 and s, numerically stabilized from. H _2_

112 Protein Dynamics: Modeling Backbone Dynamics From linear combinations of J(0), J(w N ), J(w H ), J(w N +w H ) and J(w H -w N ) the NOE, R 1 and R 2 can be calculated (Abragam, 1961.). The model fitting tries to adjust the fit parameters such that the calculated NOE, R 1 and R 2 values fit best to the corresponding experimental NOE, R 1 and R 2 values Anisotropic Modelling Fitting NOE, R 1 and R 2 of each residue to one or more anisotropic models. If is not small, say >1.2, the assumption of isotropic diffusion may not be valid and anisotropic diffusion must be assumed. In this case the relaxation rates of individual NH bond vectors depend on their orientation relative to the axes of the diffusion tensor. The anisotropic modelling requires a pdb file containing protons. The location of the pdb file must be defined via Sample in the NOE, T1 or T2 (or in all of them) methods on the method tree. Click on the Structure tab in the dialog window that appears. If you only have a pdb file without protons available use one of the existing software packages like Reduce developed by J. Michael Word in the lab of David and Jane Richardson at Duke university (Word, et. al. J. Mol. Biol. 285, , 1999). The determination of the components of the diffusion tensor and the orientation of the tensor relative to a molecular frame can be done in different ways. One way is to take the spectral density function as (Woessner, D. E., J. Chem. Phys., 37, , (1962).). The newly introduced five correlation times j are linear combinations of components (D xx, D yy, D zz ) of the diffusion tensor. The coefficients Aj also depend on D xx, D yy and D zz, and additionally on the coordinates of the NH unit vectors in the axis frame of the diffusion tensor. Since the coordinates of the NH vectors (taken from the pdb file) are typically given in another frame, they need to be rotated into the axis frame of the diffusion tensor. This rotation involves three Euler angles D xx, D yy, D zz, α, β, γ. Thus, we have a total of six unknowns (D xx, D yy, D zz, α, β, γ). From linear combinations of this spectral density function one can calculate R1 and R2 or R2/ R1 for each residue and compare it to experimental R 2 /R 1 values. A minimization of where the summation is over selected residues, R 2 /R 1 are experimental values, R 2 /R 1 are calculated and σ is the error determined for each R 2 /R 1. Selected residues means that residues probably showing conformational exchange are excluded. If the diffusion tensor is known, the above isotropic models can be rewritten by replacing the single global isotropic correlation time c with a summation of individual anisotropic correlation times. For the general anisotropic case we then get TM1 112 H _2_002

113 Protein Dynamics: Modeling Backbone Dynamics S 2 is fitted, j are calculated from the components of the diffusion tensor and are the same for each residue, Aj are calculated from components and orientation of the diffusion tensor and have to be calculated individually for each residue. Instead of M2 we get TM2 S 2, e are fitted. TM3, like TM1 but a Rex term is added to the R 2 calculation. TM4, like TM2 but a Rex term is added to the R 2 calculation. TM5 As an approximation the fast local motion is assumed to be very fast, i.e., f -> 0, fitted are S 2, S f 2 and s. Progress bars indicate the ongoing fit calculations. The GUI remains active during this time and you change the display or work on another method. Changing any data (e.g. deleting a peak) that are currently used in the calculation must be prevented. 8.6 View Results Obtained with the NOE/T1/T2 Method Once Analysis has been successfully executed View can be used to define details of the result display and get corresponding objects on screen. A tab oriented dialog window contains tabs related to general (T 1 /T 2, spectral density), R ex, S 2 and molecular displays. Individual check boxes are available depending on which calculations have been performed. In the example below the R ex tab is active and shows that Rex histograms can be shown as estimated from spectral densities or via M3 and M5 modelling. These two models contain an R ex term. Since anistropic modelling was not performed, R ex as obtained via modelling of TM3 and TM4 cannot be selected. Figure 8.7: View Dialog (Rex tab opened) to Define Details of the Result Display. H _2_

114 Protein Dynamics: Modeling Backbone Dynamics The resulting display may contain numerous display objects as shown in the following example. As usual, all objects are correlated as far as possible. If for example the cursor is moved to a certain residue above the histogram, it is also moved to this residue in all other histograms. Note that some of the items in the histograms are shown in green instead of black color. This is due to a corresponding selection of a check box in the View dialog. All residues that were selected for the global isotropic c calculation are shown in green. The second example below shows the usefulness of the combined displays. A residue shows a low T 2 value (a). The reason could be a Rex contribution. In the NOE histogram its value is relatively high (b), which means there is no fast local motion at this residue. T 1 is not low, thus T 1 /T 2 is high (c), indicating that the low T 2 value is probably not due to anisotropy. From a modelling of M1 a somewhat elevated R ex value can be seen (d). All 4 result displays seem to be consistent with each other. Figure 8.8: Example of a Result Display Showing Various Types of Objects. 114 H _2_002

115 Protein Dynamics: Modeling Backbone Dynamics Figure 8.9: Example of a Result Display Illustrating the Combined Interpretation. Apart from viewing the dynamic parameters via amino acid sequence histograms, it is also possible to display them directly on the 3-dimensional molecular structure. Again this requires a pdb file of the protein which includes protons. The location of the pdb file must be defined via Sample in the NOE, T1 or T2 (or in all of them) methods on the method tree. The 3- dimensional structure display is performed via Jmol, which is part of the Dynamics Center installation. A reference for Jmol is Jmol: an open-source Java viewer for chemical structures in 3D (see A very simple viewer that internally calls the Jmol viewer is provided. H _2_

116 Protein Dynamics: Modeling Backbone Dynamics Figure 8.10: A Simple Structure Viewer Based on Jmol. The viewer is pre-loaded with the given pdb file. A simple pull-down menu allows you to configure the display. One entry is called Execute macro. Any RasMol/Chime script can be entered here (See for example A second entry is Dynamic parameters on/off. This command presents a list of available dynamic parameters (T1/T2, spectral densities, order parameters etc.) that can be displayed as Positional Variability like for example B-factors (For a reference example see jmol.sourceforge.net/jscolors/#position%20along%20chain.). The color translation used here would correspond to the so-called relative Temperature scheme, i.e. the range between minimum and maximum value of a dynamic parameter, say J(0) is mapped into 30 colors ranging from blue (low values) to red (high values). The dynamic parameters are transferred in memory from the Dynamics Center to the structure viewer and not placed into the pdb file first. 116 H _2_002

117 Protein Dynamics: Modeling Backbone Dynamics 8.7 Report results obtained with the NOE/T1/T2 method The results generated from calculations and modeling can be very large, thus the reports may contain numerous pages. Since one of the critical issues is which of the models shall be finally accepted, back calculated R 1, R 2 and NOE values are compared to their experimental counterparts. Figure 8.11: Partial View of a PDF Report Page. The above example shows a part of a page containing the M2 modelling results. Columns for S 2, error of S 2, TauE and error of TauE are provided. In addition, there are 3 columns for dr1, dr2 and dnoe all given in percentage. The percentage is calculated from R 1 - R 1c /R 1, R 2 - R 2c /R 2 and NOE - NOE c /NOE. There are further pages that compare the AIC values of all modelling results. Usually the model with the smallest AIC is regarded as the best one, but this is a dangerous assumption since the AIC values also strongly depend on the errors of the experimental R 1, R 2 and NOE values. When the modelling does not reproduce the NOE value quite well, the overall AIC value of that model gets large, even though the more relevant R 1, R 2 and values might have been well modelled. A strategy often used is to accept models on the basis of how well they reproduce R 1 and R 2 values neglecting the NOE values. Therefore, the user may specify an expected percentage number such that if dr 1, dr 2 (as mentioned above) are below this number the corresponding model is marked with an asterisk. Independent of the AIC value this marker indicates that the particular modelling was able to reproduce R 1, R 2 well. The above example in general shows very high AIC values for M1, M3 and M5 modelling. The internal analysis showed that the main reason for this is the NOE values were not well reproduced and there were small experimental errors in the determination of the NOE values. H _2_

118 Protein Dynamics: Modeling Backbone Dynamics The results therefore need some interpretation. For example, in line 1 the AIC of model M2 has a low value (202), but it is not marked with the asterisk, whereas M3 with an AIC of 4.66e 4 is marked. In terms of reproducibility of R 1 and R 2 M3 would be the better model. Model M4 has an AIC of 6.0 which is the lowest AIC for this residue. Since the number or parameters of M4 is 3, this AIC value (AIC = c k) shows that c 2 of the fit is almost zero, which means that the fit exactly reproduced R 1, R 2 and NOE. As described in text books on statistics such perfect fits are rather unexpected and the fit is regarded as over fitting the model. But when checking the individually fitted parameters of Model M4 all of them appear to be reasonable. 8.8 Export Results Obtained with the NOE/T1/T2 Method Export provides the same information as Report in textual or Excel xls, xlsx format. 8.9 Validity of the NOE/T1/T2 Modelling When only the field strength is available the modelling of only 3 data points to models that contain up to 3 parameters causes principal and technical problems. In some cases the errors of the fitted parameters indicate that the fit is questionable, i.e. has not found the true global minimum. Resulting high AIC values may or may not indicate problems as described above. Additionally, as can be seen in the report and export result tables there are cases where the modelling failed definitively. Instead of fitted model parameters or obtained AIC values only a blank (-) is shown in the table cells. There are 3 major reasons for a failure. 1. Minimization does just not find a minimum. This is rather a rare case and can be overcome if one allows a larger number of extra minimizations with randomly selected start parameters. Approximately extra minimizations are recommended. 2. Minimization fails due to numerical problems. During minimization matrix inversions are involved which may fail. Furthermore, internal quantities like the minimum increments of parameters may be exceeded. 3. Minimization works but results don t make sense. The minimization algorithms run in an unconstrained way, but the obtained results are crosschecked with pre-defined constraints. If violations occur the minimization is regarded as having failed. Using extra minimizations often lead to more results that are accepted. The current pre-definitions are as follows: Parameter Start Value Minimum Value Maximum Value S Sf e 1x x x10-10 s 1x x x10-10 R ex 0 0 1x H _2_002

119 Protein Dynamics: Modeling Backbone Dynamics In addition, models M5 and TM5 are only accepted if Sf 2 >= S 2. In general models are only accepted if e < c if any internal correlation time is involved. If no extra iterations of start parameters are selected repeated modelling of models M1 to M5 should always yield the same results, even on different computers. Extra iterations however are based on random start parameter selections and the random numbers are obtained from the system s random generator. As a consequence repeated modelling may yield different results. If the number of extra iterations is high, say 1000, the differences of the modelling results should be small. Vice versa, if the results are different, the modelling has obviously not yet found the global minimum and is not reliable. The situation is somewhat different regarding models TM1 to TM5. These models need the fitting of the diffusion tensor which also uses the specified number of random start parameter selections if this number is larger than 500. Otherwise a minimum of 500 iterations is used. The diffusion tensor fitting may thus yield different results and the modelling of TM1 to TM5 may then also be different. Clearly, the modelling results are affected by the data quality and T 1, T 2 curve fitting. The following is often recommended: Run repetition experiments, with at least a few of the mixing times two or three times. If repetition experiments are available use the option systematic error from variance averaging (e.g. see the T1/Data/Lists tab). The error of the fitted parameters should be determined using the Monte Carlo method (see e.g. T1/Analysis) H _2_

120 Protein Dynamics: Modeling Backbone Dynamics 120 H _2_002

121 Protein Dynamics: Modelling Side Chain Dynamics 9 Protein Dynamics: Modelling Side Chain Dynamics 9.1 Introduction This method can be used to model the dynamics of CH 3 bond vectors of fractionally deuterated protein side chains. With corresponding pulse sequences it is possible to get deuterium relaxation rates originating from 13 CH 2 D isotopomers. The deuterium relaxation T 1 (D), T 1ρ (D) is encoded in triple terms of the form I z C z D z or I Z C Z D y. To get D z or D y a compensated pulse sequence that subtracts I z C z can be applied. Alternatively, one measures T 1 that relates to I z S z in a separate reference experiment and subtracts these T 1 values from T 1 and T 1ρ values obtained by the non-compensated sequence. Since the dominant relaxation mechanism for 2 H is quadrupolar relaxation, the dynamics can be described by model M2 as described in the previous chapter. If the global correlation time c is held fixed, only S2 and t e need to be fitted. At minimum a pair of deuterium T 1 and T 1ρ values is needed. The advantage of not using T 2 is that no Rex terms need to be considered in the modelling. For more details see: Lee AL., Flynn PF., and Wand AJ., J. Am. Chem. Soc. (1999) 121, Clarkson MW and Lee AL, Biochemistry (2004) 43, In the method center the method is named Sidechain Dynamics. The usage is very similar to the NOE/T1/T2 method as described in the previous chapter. Thus, it does not contain a component named Sample because the sample information is already known via the T 1 and T 1rho experiments involved. H _2_

122 Protein Dynamics: Modelling Side Chain Dynamics Figure 9.1: The Side Chain Dynamics Method is Part of the Method Tree 9.2 How to Execute the Sidechain Dynamics Method The Sidechain Dynamics method uses data and results from T 1 and T 1rho methods. Therefore, these methods must have been set-up and executed at least up to Analysis and then saved to project files (right mouse click to T1 and T1rho and Save from the popup menu). If data are available at multiple field strengths, a set of T 1 and T 1rho projects must have been saved for each field. To run the Sidechain Dynamics method, execute Sidechain Dynamics/Data and select the number of field strengths and the names of the project files (which you saved before) to be used. 122 H _2_002

123 Protein Dynamics: Modelling Side Chain Dynamics Figure 9.2: Data Dialog Window to Select Project Files to be Used for Sidechain Dynamics Modeling If non-compensated pulse sequences have been used, reference T1 experiments must have been done, analyzed and saved to a project file. The corresponding project file name must be entered in the fields named T1-ref project file. In case of compensated sequences the reference experiment is not needed and the T1-ref project file fields should be left empty or just filled with???. In practise two working procedures are possible. Alternative 1: If you start with a blank screen an automated loading takes place after the Data dialog window is closed with OK. One spectrum after the other is loaded and analyzed. If long lasting operations (peak de-convolution or curve fitting with Monte Carlo error estimation) are involved a progress bar is shown and you have to wait until the operation is finished. H _2_

124 Protein Dynamics: Modelling Side Chain Dynamics Figure 9.3: Successful loading of two Groups of T1 and T1rho Projects. [Sample (V14A* eglin c) and Assignments were provided by Andrew L. Lee (University of North Carolina at Chapel Hill)]. Alternative 2: First load the projects one-by-one (right click to T1, Open, then right click to T1ho, Open,...) and execute each method up to Analysis. This way you can better control which actions are done. If multiple instances of T 1 and T 1rho are needed on the method tree, right click for example to T1 and execute Add to tree to expand the tree. If finished then execute the Sidechain Dynamics method. This checks which data are currently on screen and offers to take them directly. After the specified projects are loaded, some internal checks are performed. All spectra should refer to the same sample and the amino acid sequence. It does not make any sense that different amino acid sequences had been defined for the T 1, and T 1rho methods. Preferably, all participating spectra contain the same number of peaks annotated the same way. However, this could be too restricting. Therefore the peak list of the first T 1 spectrum is regarded as most relevant. All peaks on the peak list are used if the corresponding peaks are also found in the other spectra. If different numbers of peaks are detected a corresponding message is presented. You may then continue with Analysis to do the modelling calculation. 124 H _2_002

125 Protein Dynamics: Modelling Side Chain Dynamics Histogram display of data vs. amino acid sequence Displaying relaxation times or other results as a function of the amino acid sequence is one of the standard tools, see e.g. section View [} 96]. In case of backbone experiments one (or no) value per residue existed. In case of sidechain experiments this is different, residues may for example have two methyl groups, thus for example two T 1 values must be shown for one sequence position. Correspondingly, the peak annotations would contain residue numbers more than once, for example the two peaks of the two methyl groups of a valine could be called Val[37]/A and Val[37]/B or V37A and V37B or anything like that. The number 37 appears twice. First a standard warning is displayed: Figure 9.4: Warning if Sequence Numbers contained in Peak Annotations are not unique. Then a display is generated in such a way that the histogram is filled from the sorted peak annotations. While the ordinal numbers at the x-axis are just running from 1.. n the actual histogram items refer to sorted peaks. In the example below a total of 28 peaks were available, the two methyl groups of valine are located next to each other, the next item (without gap in the histogram) on the right side refers to residue 37, the item on the very right of the histogram to residue 66. Figure 9.5: Example of a Histogram Display constructed from sorted Peak Annotations. [Sample (V14A* eglin c) and assignments were provided by Andrew L. Lee (University of North Carolina at Chapel Hill)]. As with other methods the Sidechain Dynamics method shows a popup menu after right clicking on it. It also contains the regular entries Open, Save, Save As, Close, Suspend and Resume. In contrary to most other methods an Add to tree entry is not active in this popup. H _2_

126 Protein Dynamics: Modelling Side Chain Dynamics The reason is simply to restrict complexity. The Save or Save As options should be used to save the current Sidechain Dynamics method into a project file on disk. As usual it contains information and settings of the method but not the modelling results itself. 9.3 Extracting Dynamic Information from T1 and T1ρ Executing Analysis is used to do the actual modelling. Figure 9.6: Analysis Dialog Window of the Sidechain Dynamics Method. The relevant model of motion to extract dynamics parameters is M2 S2, e fitted, global isotropic correlation time c is held fixed, formula given here is the numerically stabilized form. The spectral density J(ω) is related to T 1 and T 1ρ and involves the deuterium quadrupolar coupling constant, usually taken as 167kHz (Mittermaier A., Kay L.E., J. Am. Chem. Soc. 121 (45), ,).The modelling of two parameters with two data points (per field strength) is often critical. The optimize results, it is advised to redo the calculations using randomized start parameters. The number if such iterations, e.g can be given by the user. Finally, it is possible to override the errors of the T 1 and T 1ρ values obtained in the individual T 1 and T 1rho methods and assume for example that they correspond to a certain percentage of the values itself. Such a procedure may be questionable, especially if the percentage numbers are just obtained empirically. On the other side the fit parameter errors are often related to statistical errors in the data only and do not incorporate systematic errors. Then they appear to be much too small. 126 H _2_002

127 Protein Dynamics: Modelling Side Chain Dynamics 9.4 View Results The modelled S 2, e values can be display as a function of the amino acid sequence as shown in the following example. Figure 9.7: Histogram Display of Order Parameters and internal Correlation Times. [Sample (V14A* eglin c) and assignments were provided by Andrew L. Lee (University of North Carolina at Chapel Hill)]. H _2_

128 Protein Dynamics: Modelling Side Chain Dynamics 9.5 Report, Export Reports in.pdf format and export in.xls,.xlsx or.txt format can be generated with Report and Export. The example below shows part of a report. The table includes the differences between experimental and back calculated R1 an R1r values, given in percentage numbers. This gives a good impression of the quality of the modelling (here for example not very good). Figure 9.8: Part of a typical Sidechain Dynamics Report. 128 H _2_002

129 Protein Dynamics: NOE BuildUp Analysis 10 Protein Dynamics: NOE BuildUp Analysis 10.1 Introduction The NOE buildup method can be used to analyze peak intensities as a function of mixing times and derive distance information from the obtained build-up rates Sample The Sample component is identical to the relaxation methods. Some available options would probably not be used, e.g. the loading of an amino acid sequence. H _2_

130 Protein Dynamics: NOE BuildUp Analysis 10.3 Data The Data component is identical to the relaxation methods. However a standard pulse sequence that would generate a pseudo 3D spectrum is not available, typically a series of 2D spectra are measured and the mixing times in units of seconds have to be entered manually. As the peak picking method the threshold based peak picking is recommended. After the spectra are loaded use the mouse to define a signal free region from which a peak picking threshold is derived. The picked peaks can be cleaned up using the options from the peak popup menu. Figure 10.1: Example of a Threshold based Peak Picking of a NOE Spectrum. 130 H _2_002

131 Protein Dynamics: NOE BuildUp Analysis 10.4 Analysis The Analysis component is used to fit the peak integrals to particular build-up functions and extract the relevant parameters. It can be executed if the Data component was already successfully executed before. Figure 10.2: Analysis Dialog Window of the NOE build-up Method. Two approximations are offered to fit NOE build-up curves: Quadratic approximation: f(t)=a * t + b * t 2 Bi-exponential build-up f(t) = a * e-b * t * [1 e b * t] The fit functions are indicated in the dialog window. The variable t is the mixing time. The distance information between nuclei is related in the cross-relaxation rates which in turn relate to the formal fit parameters a, b and c. In case of the quadratic approximation -b/a is evaluated, in case of the bi-exponential build-up c/2 is evaluated. As always the error calculation of the fit parameter errors can be selected, using error estimation by weighted fit is recommended in most cases. The confidence level expresses the probability that a fit value is within given range. Depending on the percentage number the value range fit +/- error is increased or decreased. The larger the confidence the larger the range. H _2_

132 Protein Dynamics: NOE BuildUp Analysis 10.5 View The view component allows viewing of the obtained results. The dialog window is customized to the necessary items, e.g. histogram displays related to the amino acid sequence are not offered. A typical result view looks as follows: Figure 10.3: View Results after fitting NOE build-up Curves. The fit curve display also indicates the calculated distance in Angstrom and for security also the global correlation time which influences the display calculation as well. It should be set correctly at the beginning, see Sample [} 129] Report, Export These two components are more or less identical to other methods and described in previous sections. In general, the output contains the calculated distances not only the fit parameters a, b, c. 132 H _2_002

133 Protein Dynamics: S/N Comparision 11 Protein Dynamics: S/N Comparision 11.1 Introduction The S/N of a spectrum acquired with Band Selective Optimized-Flip-Angle Short-Transient ( SOFAST) methods depends on T 1, pulse angle, recycle delay and number of scans. The signal dependency of SOFAST HMQC experiments is described as (Schanda P. et al., J. Biomolecular NMR, 33(4), , (2005)). T rec is the total time (recycle delay + pulse length + acquisition time) n, T scan is the acquisition time per scan, β is the pulse angle and T 1 is the longitudinal relaxation time. With fixed n, T scan and β one may optimize the S/N by variation of the recycle delay Sample The Sample component is identical to the relaxation methods Data The Data component allows you to select a series of 2D spectra acquired with different recycle delays D1. The handling of peaks and integrals is identical to the relaxation methods described in Protein Dynamics: Basic Relaxation Analysis [} 87] Analysis The Analysis component is used to calculate the signal-to-noise of all peaks in the spectra and to normalize it with respect to the recycle delay D1 or the sum of acquisition time and recycle delay. Figure 11.1: Analysis Dialog Window of the SN Comparison Method. The noise is the standard deviation of a signal free region given in ppm. Peak intensities or peak integrals are used depending on the selection in Data. H _2_

134 Protein Dynamics: S/N Comparision 11.5 View The View component allows you to view the obtained results. A typical result view looks as follows: Figure 11.2: View Results after S/N Calculation as a Function of D1. The fit curve display also indicates the calculated distance in Angstrom and for security also the global correlation time which influences the display calculation as well. It should be set correctly at the beginning, see Sample [} 133] Report, Export These two components are more or less identical to other methods and described in previous sections. 134 H _2_002

135 Protein Dynamics: CEST 12 Protein Dynamics: CEST 12.1 Introduction CEST (Chemical Exchange Saturation Transfer) serves to detect (invisible, low populated) excited protein states that are in slow exchange with visible ground states. Various regions in a spectrum are irradiated with a weak B 1 field to observe effects on ground state peaks. For more details see Vallarupalli P., Bouvignies G., and Kay L., J. Am. Chem. Soc. (2012) 134, The CEST method is implemented among the protein dynamics methods. Figure 12.1: CEST is Located in the Protein Dynamics Method Tree 12.2 Sample The Sample component is used to provide some information on the sample. This may include general information but also the amino acid sequence of the protein which is later used for displaying results as a function of sequence, see View [} 138]. H _2_

136 Protein Dynamics: CEST 12.3 Data The Data component is used to select a CEST spectrum and define details about peak picking and peak integration. If the Bruker released pulse program hsqc_cest_etf3gpsitc3d (TopSpin 3.5 or later) is used, pseudo 3D spectra are generated with TD in F1 equals number of planes (= number of irradiation frequencies), NUC1 in F1 is for example set to 15 N and additionally used parameter files are cpdprg8: 90x_240y_90x, cpdprg7: cwp. Further relevant parameters are D21, which equals the irradiation time T ex of the B 1 field (e.g. 0.4s) and CONST9, which equals the B1 field strength (e.g. 20 Hz). A frequency list (parameter FQ3LIST, for an example see exam_15n_cest) of the following type is generated: bf ppm Here, bf means that the following numbers are absolute values and ppm indicates the units. The list contains TD-1 entries, the first experiment (corresponding to T ex = 0) is used to normalize all other planes in the pseudo 3D spectrum and is recorded in addition to the list entries. All other known formats of such lists, e.g. sfo or O instead of bf and Hz instead of ppm are also supported. To get the correct frequency list taken by the Dynamics Center, the parameter NUC1 must be set correctly in acqu3s, e.g. ##$NUC1= <15N> Analysis Peak intensities or peak integrals available as a function of B 1 irradiation frequency can now be fit to In case of exchange between a ground state G and an excited state E with zero equilibrium magnetization, R is defined as -R 2G -k GE -w G w 1 k EG 0 0 w G -R 2G -k GE 0 0 k EG 0 -w 1 0 -R 1G -k GE 0 0 k EG k GE 0 0 -R 2E -k EG -w E 0 0 k GE 0 w E -R 2E -k EG k GE -w 1 0 -R 1E -k EG 136 H _2_002

137 Protein Dynamics: CEST The relations between populations and exchange rates are given by pg = keg/kex, pe = kge/kex and kex = kge+keg. Furthermore, it was assumed that R 1G = R 1E. Fitted parameters are k ex, p E, w G, w E, R 1G (=R 1E ), R 2G, R 2E, I o. Since all profiles are normalized, I o is close to 1.0. Details of the fit can be specified in a dialog window. Figure 12.2: Dialog Window for CEST Analysis, Systematic Variation of Start Values for w E Activated To obtain good fits, reasonable start parameter values are needed. The value for w E is different from peak to peak and sometimes hard to determine in the CEST profiles. Therefore, it is recommended to activate the try different we start parameters option. Start values for w E are then systematically varied over the complete profiles. Since computing times increase, it also recommended to use the activate parallel processing feature. If the computer contains N processors, N-1 will be used for the CEST fit. Under the start parameters tab further values can be given. Figure 12.3: The Start Parameters Tab H _2_

138 Protein Dynamics: CEST In this example the value for w E is disabled since it will be varied systematically anyway. The w G value is also disabled because it gets determined from the profiles of each peak. If this automatic determination fails a default value is supplied. Suitable average values for k ex, p E, R 1G (=R 1E ), R 2G and R 2E can normally be given by the user, I o should be close to View Similar to all other methods View is used to define the details of the result viewing for each peak. Figure 12.4: Successful Fit of the CEST Profile Closest to the Cursor Report Report is used similarly to all other methods to generate a pdf report of the current analysis Export Export is used similarly to all other methods to save data and results in textual of EXCEL format. Included are both the original peak integrals as well as the normalized value. 138 H _2_002

139 ProteinDynamics: Multi T2 / Kd 13 ProteinDynamics: Multi T2 / Kd 13.1 Introduction The methods described so far use a set of spectra (either stored as pseudo spectra or a real series of spectra) as a function of a variable (time, gradient strength etc.) to determine quantities such as relaxation times, diffusion constants, rate constants etc. Such quantities itself can be taken as a function of a further variable, e.g. temperature. The current Multi T 2 /K D method offers one such case. T 2 relaxation experiments of protein-ligand mixtures are taken at different ligand concentrations while the protein concentration is kept constant. If the ligands bind to the protein in the fast exchange regime, the observed linewidths of the ligand signals are a superposition of linewidths of free ligands and bound ligands. These data can then be used to obtain the dissociation constant K D, see e.g. Supporting information, Trevino et al., /pnas Sample The Sample component can be used to specify some simple information about the given sample. It is only used for documentation purposes and may be contained in reports or exports Data The Multi T 2 /K d method only uses series of pseudo2d spectra. In the Data dialog the user can provide the spectra names and associated concentration values (in mm) of the ligand. The first T 2 spectrum must be the pure ligand spectrum, i.e. not containing any protein. Formally the ligand concentration should be set to 0 even though it does not have any real meaning. The other tabs (Peaks, Integrals, Lists) are identical to the ones used in the regular T 2 method. The first pseudo2d spectrum is loaded and peak picked accordingly, the first trace of this spectrum is displayed together with a slider that allows you to navigate to other traces of the first spectrum. H _2_

140 ProteinDynamics: Multi T2 / Kd Figure 13.1: The Multi T2/Kd Data Dialog See also 2 Drag & Drop of Spectra to the Method Tree [} 26] 2 Data [} 40] 13.4 Analysis The analysis of each pseudo2d spectrum is identical to the analysis of a single T 2 spectrum as described in chapter 6.2: The T1, T2, T1rho Methods. For each of the peaks in the first pseudo2d spectrum T 2 constants are calculated based on the selected fit function. In principle one could apply a multi component fit (e.g. number of terms = 2) and if such fits succeed and are statistically better than a single component fit, the T 2 values of the second term would be used but to avoid mixtures of single and multi component results, we recommend to set number of terms = 1. The Analysis dialog window contains a further tab named Kd Analysis. The formula used to fit R 2b and K d is shown. To follow the literature R 2 (=1/T 2 ) values are used. R 2obs are the observed linewidths in the different spectra. The formula also involves the number of binding sites for the ligand and the protein concentration which must be both given. R 2f is calculated from the first pseudo2d spectrum which corresponds to the free ligand, i.e. does not contain any protein. An additional option allows you the discard mixing times below a certain value during the determination of the individual T 2 values of each peak in the individual pseudo2d spectra. The reason is that the T 2 fits sometimes get better if inaccurate data points at very small mixing times are skipped. 140 H _2_002

141 ProteinDynamics: Multi T2 / Kd Figure 13.2: The Multi T2/Kd Analysis Dialog Contains a Tab Named Kd Anlaysis Once the first spectrum is analyzed its peaks are projected to the other pseudo2d spectra, local peak snapping is automatically applied and the calculation of the T 2 values is repeated for all peaks in all the other spectra View This component is used to specify details of the fit curve display and is mostly identical to the other methods. When moving the mouse pointer to a peak (related to a certain trace in the first spectrum), the T 2 curves from all available pseudo spectra are automatically cumulated in a window (upper right window in the example below) and the R 2obs values as a function of the ligand concentration are shown in a second window (lower right in the example below). From this curve K d is determined. H _2_

142 ProteinDynamics: Multi T2 / Kd Figure 13.3: Result View of the Multi T2/Kd Method Depending on the selected options further details may be shown, e.g. the residuals of the T 2 fit curves Report, Export These two components are similar to Report, Export in other methods and serve to create PDF reports and exports in.xls, xlsx or.txt format. Currently, the report also includes the individual T 2 fit results of each peak of each of the pseudo2d spectra while the export only includes data and results of the K D fit. Instead of fitting the full R 2obs curve as a function of the ligand concentration one might estimate K D by just using two experiments as this would save a lot of experiment time. In this case one would however have to specify the (temperature dependent) rotational correlation time of the protein and derive an estimate of R2b. Currently, this is not offered in the reports or exported data. 142 H _2_002

143 Protein Dynamics: Interfacing to Relax 14 Protein Dynamics: Interfacing to Relax As described in Protein Dynamics: Modeling Backbone Dynamics [} 103] the modelling of protein backbone dynamic parameters from relaxation parameters is done such that residues with no fast internal mobility and no relaxation exchange are selected via their NOE and T 1 /T 2 values. From these residues a global isotropic correlation time or the diffusion tensor are derived. Both are kept fixed when modelling M1--M2 and TM1--TM5. This strategy is simple and is still used by many software packages in this field. Figure 14.1: Standard Export dialog Window of any Relaxation Method. The software Relax (d'auvergne, E. J. and Gooley, P. R., J. Biomol. NMR, 40(2), , 2008) (uses a more advanced modelling scheme. Relax (Relax version or later for Windows can be downloaded from A Linux version can be downloaded from is able to read the Dynamics Center output of the NOE, T 1, T 2 and R ex methods. To use these methods with Relax, execute these methods as described in Protein Dynamics: Interfacing to Relax on page 189 [} 143] and use Export to generate individual output files in textual form. The output file names are arbitrary, for example NOE.txt, T1.txt, T2.txt. It is recommended to activate all check boxes in the export dialog to get complete outputs. Relax scans the files and extracts relevant parts automatically. If data in multiple fields is available, analysis and export have to be repeated at each field strength using different output file names. Relax, which uses the T 1, T 2 and NOE data, is used in combination with NESSY, an open source (GPL) software. NESSY additionally analyses relaxation dispersion experiments at one or multiple magnetic fields. Data is individually fitted to different 2 state, 3 state and n- state models, and the model selection is performed according to Aikaike information criteria (AIC), AIC using second order correction for small sample size (AICc) or F-test. H _2_

144 Protein Dynamics: Interfacing to Relax Figure 14.2: NESSY Dialog Window to Import Output from the Dynamics Center. The quality of the modelling depends on the quality of the NOE, T1 and T2 values. The following recommendations regarding the T1 and T2 data are given for best results obtained with relax: If repetition experiments are available, use the option systematic error from variance averaging (for example the T1/Data/Lists tab). The error of the fitted parameters should be determined using the Monte Carlo method (for example T1/Analysis). 144 H _2_002

145 Time Domain Dynamics Method Center 15 Time Domain Dynamics Method Center 15.1 Introduction The Time Domain Dynamics method center is used for the analysis of NMR data in the time domain. It is used exactly the same way as in all other applications of the Dynamics Center software. This means that each method consists of components Sample, Data, Analysis, View, Report, and Export. These components have to be executed in this order. After a successful execution the corresponding nodes on the method tree are green color-coded. A red color-code indicates that a component is currently executing, e.g. doing a longer calculation. Any subsequent components can only be executed if the previous component is shown in green. In the example above, state A shows a method just opened. State B indicates the method is opened and Sample and Data have been successfully executed, while Analysis is currently in progress. The subsequent View can only be executed after analysis has turned green. However, it is possible to execute components of other methods in parallel. It is also possible to right click to a method and get useful options in a popup menu, such as Open, Close, Save (projects), Suspend, Resume, Add to tree. These serve for more efficient work, or simplification of the display or for cloning methods to allow a parallel analysis of different data. Please refer to chapter 5.2 for an introduction. The Time Domain method Center offers the method TimeDomain1D which is used for the determination of T 1, T 2, T 1rho times, but also for curve fitting according to multicomponent models as encountered in polymer science or multiphase materials for example. A second method named TimeDomain2D can be used to apply inverse Laplace techniques to 2D data such as T1-T2 or Diffusion-Diffusion and obtain characteristic distribution maps of corresponding physical parameters such as relaxation times or diffusion constants. To run this method an extra license is required. H _2_

146 Time Domain Dynamics Method Center Finally, there are the methods qsrc and qsrc(3) which serve for the quantification of components in mixtures such as crystalline and amorphous components in a solid tablet. Since the quality of the quantification results improves strongly with recent hardware developments, it is strongly advised to contact application experts (aicapps@bruker.com) to make sure that your hardware is suitable The TimeDomain1D Method Sample The Sample component is used to provide information about the sample and is only used for report purposes (see Report [} 158]) Data The Data component is used to define the details related to the input files (.dps or.sig). 146 H _2_002

147 Time Domain Dynamics Method Center Input files are.dps files (Data Point Storing) and.sig files as generated by Bruker s minispec. They can be located via the browse button. A field called number of lines to be skipped allows the user to skip any number of lines preceding the actual data in a.dps or.sig file, e.g. an optional header section. In the example below the header section includes the first seven lines. Figure 15.1: Example of a Header Section in a dps File. Alternatively, this option can be used to discard the initial data points. This might make sense for cpmg data acquired with the Bruker minispec ProFiler ( mr/td-nmr/minispec-profiler.html), where simultaneous T 1 and T 2 effects complicate the evaluation of the first few echo intensities [M.D. Hürlimann, chapter 3 in Single Sided NMR, Eds F. Casanova et al, Springer-Verlag Berlin Heidelberg 2011]. Similarly, the field define comment sign allows the user to disregard lines which start with the symbol as defined here. Note: With Bruker standard applications comment signs are not used, but they can occur for customized NMR experiments. Standard data files are available in two columns format (x, y) or three columns format (index, x, y). Proper column numbers, i.e. 1 and 2 or 2 and 3, must be specified to get x (time) and y (amplitude) values loaded correctly. The following example shows a.dps file with two columns, thus the correct settings are 1 for x values and 2 for y values. H _2_

148 Time Domain Dynamics Method Center Figure 15.2: Example of a dps File Containing Time Values in Column 1 and Amplitudes in Column 2. The next example shows a.dps file with three columns, thus the correct settings are 2 for x values and 3 for y values. Figure 15.3: Example of a dps File Containing Time Values in Column 2 and Amplitudes in Column 3. The.dps files generated with Bruker standard applications contain the columns tab separated. If the check box use all available data points is checked, then all data points are displayed and used for analysis. In certain cases it can prove beneficial to truncate the recorded relaxation curve. First and last time point defining the range to be used must be given in this case. Note: In the case of FID measurements, the decay after the first few hundred microseconds is dominated by magnetic field inhomogeneities of the instrument rather than characteristic sample properties Analysis The Analysis component is used to fit the intensities to particular functions and extract relevant relaxation parameters. It can be executed if the Data component was already successfully executed before. 148 H _2_002

149 Time Domain Dynamics Method Center Figure 15.4: The Analysis Dialog Window Used for Time Domain 1D data The Analysis window consists of four tabs named Fit Functions, Start Parameters, and ILT (Inverse Laplace Transform) and ILT regions which are described in more details in the following sections. H _2_

150 Time Domain Dynamics Method Center Fit Functions The upper part of the window under Fit Functions contains a small library of predefined functions that are encountered frequently in TD-NMR. The lower part contains the option to fit the experimental data to multiple components if Exponential decay, Lorentzian decay or Gaussian decay is selected. E.g. for number of components = 3 the equation for Exponential decay is: I(t) = I 0. e t/t0 + I 1. e t/t1 + I 2. e t/t2 In such cases the software actually tries fewer components as well and presents the best solution by evaluation of the so-called AIC criterion. Example: Exponential decay with three components is selected. Then the fit is tried with one, two and three components. If all succeed, the solution with the lowest AIC value is presented. If however the tic box prioritize highest number of function components is activated, then the AIC criterion is not evaluated but the result with three components is accepted if the fit succeeded. If not, the result with two components is accepted if the fit succeeded otherwise the fit with one component. It is also possible to fit a constant offset to the selected function in which case a term C is added to the respective equation, e.g. for Exponential decay: I(t) = I 0. e t/t + C It should be mentioned that such an offset should have physical evidence. A truncated decay that has not yet reached zero intensity is for example not a real offset. Adding the fit of an offset parameter in this case would lead to a wrong fit of the decay constant. In order to provide fitted parameters within a confidence interval a confidence level needs to be given. The confidence interval of a fitted parameter is then calculated by multiplying the error of the fitted parameter with a factor taken from the inverse of the Student-s-T cumulative distribution at a given confidence level and a number of degrees of freedom. The confidence interval calculates as fitted parameter +/- fitted parameter error. The larger the chosen confidence value, the larger the interval. Predefined Experiments Experiments that yield data which can be analyzed with the predefined fit functions are delivered by Bruker and can be found in the MiniSpec installation folder. Note: User defined project files should not be saved in this directory as it is deleted upon reinstallation or uninstall of the software. 150 H _2_002

151 Time Domain Dynamics Method Center Exponential Decay Exponential decays are observed frequently for liquid samples. Multi-exponential decays are often observed for multi-phase liquid samples for example emulsions and for liquids (water or oil) within pores of porous materials. In this case the tab ILT is active. NMR experiment: t2_cp_mb.app, fid_cp_mb.app, fid_mb.app Lorentzian decay For many complex materials the FID cannot be described by either an exponential or Gaussian shape. NMR experiment: fid_mb.app, fid_cp_mb.app Gaussian decay Gaussian decays are often observed in solids, semi-solids or materials that show a restricted mobility on a molecular level. NMR experiment: t2_se_mb.app, data_ow4.app, fid_ow4.app Gaussian-Exponential decay Complex materials often yield a signal which is a superposition of more than one domain. An example is a polymer material which also contains solvent, unpolymerized monomer or small molecule additives. Often the signal stemming from the polymer can be fitted with a Gaussian term and the solvent with an exponential term. NMR experiment: t2_se_mb.app T1 Inversion recovery T 1 inversion recovery experiments are used to determine the spin-lattice relaxation time (T 1 ) of a nucleus. NMR experiment: t1_ir_mb.app T1 Saturation recovery T 1 saturation recovery experiments are used to determine the spin-lattice relaxation time (T 1 ) of a nucleus. In this case the tab ILT is active. NMR experiment: t1_sr_mb.app Abragam-Weibull-Exponential This function is commonly used to describe the FID decay of semicrystalline polymers such as polyolefins. The Abragam term describes the decay of crystalline materials [A. Abragam, The Principles of Nuclear Magnetism 1961, Oxford Univ. Press], the Weibull (or strechted exponential) term describes the interfacial domain between crystals and amorphous regions, and the exponential term describes the decay stemming from the more mobile amorphous domain. The amplitudes A 0, A 1, and A 2 represent the crystalline, interfacial and amorphous domain respectively. This model only applies for measurement temperatures sufficiently above the glass transition temperature of the polymer! [D. Daydali, A. M. Kenwright, Polymer, 1994, 35, ; E. H. Hansen, Macromolecules 2009, 42, ; K. Saalwächter, Macromol. Chem. Phys. 2006, 207, ; B. Blümich, V. M. Litvinov, Macromolecules H _2_

152 Time Domain Dynamics Method Center 2007, 40, ; V. M. Litvinov chapter 11 in NMR Spectroscopy of Polymers: Innovative Strategies for complex Macromolecules, Cheng, H. et al.; ACS Symposium Series, American Chemical Society: Washington D.C, 2011] NMR experiment: fid_mb.app, fid_cp_mb.app Gauss-Gauss-Exponential This function is similar to the Abragam-Weibull-Exponential function and can be used alternatively. NMR experiment: fid_mb.app, fid_cp_mb.app Cross-link density I This function is used to describe the decay observed for rubber and elastomer materials. The first term describes the decay due to restricted motions of polymer chains between crosslinks. q is a dimensionless motion factor typically on the order of 10-4 to The more rigid (highly crosslinked) a material is the higher q. This function is implemented in the data analysis of the Bruker standard application t2_se_cld_nr.app in the minispec software. M is the second Moment and equals 0.86 * for natural rubber. The underlying theory is described by Kuhn and Schneider and others [W. Kuhn, H. Schneider Macromolecules 1994, 27, ; H. Schneider Polymer Bulletin 1996, 37, ]. NMR experiment: t2_se_mb.app, t2_se_cld_nr.app Cross-link density II This function is similar to Cross-link density I and can be used alternatively. Here the residual dipolar coupling (D res ) is dependent on the rigidity (cross-link density). The theory and applications in the field of rubber and elastomer materials are described for example by Saalwächter et al. [K. Saalwächter Rubber Chemistry and Technology 2012, 85, ]. NMR experiment: t2_se_mb.app, t2_se_cld_nr.app For more details or assistance please contact Helpdesk Europe ( ), Helpdesk USA ( ) or minispec.sls@bruker.com Finally, there is a function named User defined (python style). This can be used if other experiments are performed or data shall be modelled with non-predefined functions. 152 H _2_002

153 Time Domain Dynamics Method Center Start Parameters Depending on the selected fit function an automated estimation of start parameters is offered otherwise they must be given by the user. If a function with multiple components has been selected, start values must be provided in comma separated form. It is advised to save work into a project file (right click on TimeDomain 1D, Save or SaveAs from the popup) after the analysis has been performed. This way the start parameters are available the next time this project is loaded (right click on TimeDomain 1D, Open from the popup). If the option User defined (python style) is chosen then any function can be entered. In this case the number of fit parameters, comma separated start values and comma separated unit must be provided as seen in the following examples: Function Number of fit parameters Start values Units y=exp(-p[0]*x) s y=p[0]*exp(-p[1]*x) , 5.0 %, s y=p[0]*exp(-p[1]*x)+p[2] , 5.0, 10 %, s, none The notation always starts with y=, the function variable is called x and parameter are named p[0], p[1], p[2], If a fit parameter does not have a real unit, just supply the keyword none ILT (Inverse Laplace) An alternative to fitting a selected number of function components is the application of an inverse Laplace transform (ILT). Using time domain data this technique is commonly used in the analysis of relaxation data, currently however only to exponential decay, saturation recovery and full inversion recovery. The key of ILT is to explain the data by a linear combination of terms evaluated at fixed points. The user only needs to carefully specify the expected range of relaxation times. The minimum and maximum expected numbers should be extended a bit (e.g. 20%) to avoid numerical artifacts during the calculation. For other settings the default values are usually good to use, especially grid type = log grid, regulariztion = second derivative, find alpha automatically. For the interested user mathematical details can be found in S. Provencher, Computer Physics Communications, 27, , (1982). H _2_

154 Time Domain Dynamics Method Center Figure 15.5: Dialog Window to Setup an ILT Calculation for Time Domain Data. ILT is computational expensive, can be instable and yields larger errors of fitted parameters. By choosing a proper number of grid points (e.g ) and by activating apply data compression calculation times can be reduced and results can be improved. Data compression here means that the so-called kernel matrix spanned by the number of data points and number of grid points undergoes an eigenvalue analysis and only significant eigenvalues are kept. This number, typically 8-15, can either be given by the user or 0 can be supplied in which case the program keeps all eigenvalues larger than 1.0e-5 which typically results in 8-10 eigenvalues. Note: Especially for time domain data with a large number of data points such as FIDs or cpmg decays the option apply data compression is highly recommended. Also the number of grid points influences the calculation time, usually more than 150 data points are not needed. The regularization option Thikhonov is considerably faster, but tends to yield sharper distributions of relaxation times. For more information please consult the tutorial Time Domain Dynamics (see Time Domain NMR). 154 H _2_002

155 Time Domain Dynamics Method Center ILT regions For some applications that use ILT for modelling the data, it may be relevant to define regions in the ILT profile, integrate these regions and calculate ratios. The ILT regions tab serves to activate this. Figure 15.6: The ILT Regions Tab is Part of the Analysis Dialog Window Up to 8 regions can be specified, each with left and right time values in seconds. To activate the calculation check the tic box named perform region analysis View The View component allows viewing the obtained results. It is environment-dependent, i.e. its options depend on the kind of Analysis that was carried out. If, for example, ILT was not calculated then the corresponding tab (ILT plot) does not appear. Figure 15.7: View Dialog H _2_

156 Time Domain Dynamics Method Center Show fit curve in separate internal window means that a new display window is created in the main window and filled with fit the curve. Additional global cumulate opens a further display window and adds the fit curve there as well. This additional window is also open to other projects and thus allows you to add different fit curves of different data into the same window. Display residuals shows the difference between experimental and fitted data. These differences should be randomly distributed. A statistical 2-sided Shapiro Wilk test is available via properties of that plot. If the cumulated fit curve tic box is switched on residuals are also cumulated. If the selected fit function permits it, a logarithmic y axis can be selected. Experimental points (black), continuous fit curve (blue line) and possibly obtained points from ILT (red crosses) can be shown. Figure 15.8: Typical View after Time Domain Analysis The ILT plot tab is used to switch a ILT profile plot on. It shows for example the distribution of relaxation times needed to explain the data. If ILT regions have been defined and activated under Analysis, these regions are indicated graphically on the profile. They are automatically integrated and normalized, the results are visible under Properties from the popup menu of the profile and are also included in Report and Export. A typical result looks like: Integrated regions Region 1: left = right = integral = % Region 2: left = right = integral = % Context Sensitive Pop-Up Menus Context sensitive popup menus are available after a right mouse button click. Options are: Toggle full display This is the standard display toggle used to show the fit curve in full screen. 156 H _2_002

157 Time Domain Dynamics Method Center Visibility Controls the visibility of individual display objects. This is only used occasionally since there are several other ways to customize the display (Toggle, View, Suspend on the method tree). Add external data This special option allows you to read a (csv) text file that contains comma separated x variable values and y values (without units) in each line, like: , , , These are loaded and displayed together with the fit curve display on screen. Individual color and line style can be selected. The main intention of this tool is to compare, for example, simulated data with experimental data. The relative y-scaling is preserved, in order to get proper displays the external supplier of the csv file should scale the data properly. The range of x variable values should essentially correspond to the range on display. Export graphics Allows export of the current window as a graphic in common formats (jpg, bmp, png, emf, svg, eps). Properties Some properties of the current fit curve are shown. These include the used x and y values, fit results, and goodness of fit. Close Closes the respective window. This option is not available for the window displaying the experimental data. Figure 15.9: Typical Display of analysis results of Time Domain Dynamics H _2_

158 Time Domain Dynamics Method Center The above window shows a typical display of the analysis results of the Time Domain Dynamics Method Center with the experimental data in the top left, fit curve in the top right, residuals in the bottom left and ILT distribution in the bottom right Report A PDF report that contains loaded information, as well as results, can be generated in PDF format. Graphical components include the current experimental decay curve and fit curve displays. Numerical components include the sample, decay and fit information. The user may select desired components in a dialog window. The name of the PDF file must also be specified. At the end of the report generation AcroRead is automatically launched to display the report. Some versions of AcroRead do not display the PDF file if it was not specified with its absolute path name, e.g. c:\tmp\test.pdf. If you don t have AcroRead on your computer an alternative PDF display program may selected under config/preferences/default PDF Viewer Export Information, especially fit information and fit results, can be exported to a file on disk. The supported formats are text and xls (or xlsx). This allows users to present data with standard tools, e.g. chart diagrams in EXCEL or to load decay curves into other software packages to repeat the fit calculations, or fit other functions not offered in the Dynamics Center The TimeDomain2D Method Sample The Sample component is used to provide information about the sample and is only used for report purposes (see Report [} 163]) Data The Data component is used to define the input files which are all stored in textual form. 158 H _2_002

159 Time Domain Dynamics Method Center Figure 15.10: Data Dialog of the TimeDomain2D Method When hovering the mouse pointer over the edit fields tooltips are shown. The 2D text file should contain n1 lines with n2 values each. The values may be separated by comma, semicolon, blanks or tabs which is automatically detected. The other two text files contain the so-called x variable values in both dimensions, in both cases just one number per line. They are generated when running the corresponding experiments on Minispec. Quite often the number of variables is very different and not limited. In the examples below there are values in the F1 dimension (vertical, T 2, CPMG sequence) and only 30 values in the F2 dimension (horizontal, T 1, inversion recovery sequence). The type of experiment is not limited to T 1 or T 2 but can be any combination of T 1, T 2 and Diffusion. Details are selected under the Dimensions/Kernels tab. Figure 15.11: The Dimensions/Kernels Tab of the Data Dialog The so called kernels are shown in red color. These are quantities internally calculated and used during the data modelling. The variables are always indicated by x, depending on the experiments they are either mixing times (T 1, T 2 ) or B values (Diffusion). H _2_

160 Time Domain Dynamics Method Center Analysis The Analysis component is used to model the 2D input data relaxation parameters. It can be executed if the Data component was already successfully executed before. Figure 15.12: The Analysis Dialog Window Used for Time Domain 2D data The data modelling is based on the so-called FISTA algorithm as described in P.D. Teal and C. Eccles, Inverse Problems, 31(4):045010, April It compares well to 2D ILT as published in Venkataramanan L., et al., IEEE Transactions on signal processing, Vol 50, N. 5, May Referring to the T 1 -T 2 example above a so-called 2D grid is defined by minimum and maximum expected values for T 1 and T 2 and the number of grid points in both dimensions. The goal of the data modeling is to describe the data by a linear combination of kernel expressions taken at each grid point. Since the needed cpu time and numerical stability depend on the number of grid points, they should be chosen carefully, i.e. conservatively low. If the grid spans different orders of magnitude a log spacing of the grid points is recommended. Internally a so-called Tikhonov regularization is applied. The strength of this regularization (often called alpha in the literature) must be specified, 1.0 is suggested as an initial value. An automated determination of the best alpha is currently not offered. In contrary to the 2D ILT as published by Venkataramanan et al., the FISTA algorithm does not require an explicit kernel compression to achieve acceptable computing times. Nevertheless this is offered here. It does not speed up the calculation but may lead to more stable results. If the option is selected we recommend to set the number of eigenvalues to 0 which means that the software automatically finds a reasonable number by itself. If a number larger than 0 is specified, it will be taken. 160 H _2_002

161 Time Domain Dynamics Method Center As in the case of ILT applied in time domain 1D analysis one may be interested to integrate regions in the obtained grid amplitudes. The ILT regions tab allows you to define up to 8 2D regions. Figure 15.13: ILT Regions Tab to Define 2D Regions For each region bottom-top and left-right values must be given. These regions will be integrated as part of the ILT analysis and included in report and export. They will also be shown graphically on the 2D ILT result display, see View below View The View component allows viewing the obtained results, i.e. a 2D display of the obtained grid amplitudes that contributed to the best solution. This map shows the distribution of physical quantities such as T 1 and T 2 and is characteristic for the sample. The limits of the display would usually be the same as used for the calculation. H _2_

162 Time Domain Dynamics Method Center Figure 15.14: The View Dialog Window The number of points used for the display can be higher. However, since the display is generated by a sum of Gaussian 2D shapes at each non-zero grid amplitude, an unwanted artificially increased resolution may come up. This can be compensated by increasing the minimum width during the simulation, e.g. to 2 or 3 points. Best results are often obtained if one keeps the number of displayed points similar to the ones used for the calculation and let the minimum width stay at 1.0. Figure 15.15: Typical Result Display of a FISTA Calculation 162 H _2_002

163 Time Domain Dynamics Method Center When moving the cursor in the 2D display some tools are available on a popup menu by clicking the right mouse button. If 2D regions have been defined under Analysis, they will be shown graphically. It is possible to move or resize them with the cursor. In this case the new region borders are internally transferred to the region table. Since an already done analysis, report and export are now possibly inconsistent the status on the method tree changes and you have to restart at Analysis again. Figure 15.16: Interactive Change of Regions Causes Change on Tree Report A PDF report can be generated. Essentially, the data loaded, used calculation parameters and the obtained 2D map of grid amplitudes are included in graphical form. The user may select desired components in a dialog window. The name of the PDF file must also be specified. At the end of the report generation AcroRead is automatically launched to display it. Some versions of AcroRead do not display the PDF file if it was not specified with its absolute path name, e.g. c:\tmp\test.pdf. If you don t have AcroRead on your computer an alternative PDF display program may be defined under config/preferences/default PDF Viewer Export Information, especially fit information and fit results, can be exported to a file on disk. The supported formats are text and xls (or xlsx). This allows users to present data with standard tools, e.g. chart diagrams in EXCEL. H _2_

164 Time Domain Dynamics Method Center 15.4 The qsrc Method As stated in Stueber D. & Jehle S., J Pharm Sci Jul;106(7): doi: / j.xphs Epub 2017 Apr 12. characterization and quantification of APIs (active pharmaceutical ingredients) play a central role in the pharmaceutical industry. The qsrc method allows you to quantify compounds in solid mixtures and is based on the evaluation of T 1 saturation recovery (SRC) data measured on bench top time-domain NMR instruments. A patent describing the method has been filed (US provisional patent filing (Appl. No. 62/294,395) 2/12/2016)). The T 1 SRC data of the mixture are a superposition of the T 1 data of the individual compounds, i.e. These data could be modelled by a multi-component fit to obtain the individual contributions A i. In practice this type of modelling may suffer from the fact the A i are highly correlated, in fact, the formula can be re-arranged to contain only a single coefficient A being the sum of all Ai. Alternatively, one may also measure the T 1 SRC data of the individual compounds of the mixture in separate experiments (once) and then combine all data as: This means that each data point S(t k ) of the mixture is given as a linear combination of the data points of the individual compounds taken at the same time point t k. The data modelling then yields the coefficients c i. The following picture illustrates the case of 2 compounds. 164 H _2_002

165 Time Domain Dynamics Method Center Figure 15.17: Principle of the Data Modeling of a 2-Component System In order to get correct quantification, the SRC curve of the individual compound-i needs to be properly scaled by: nh i is the number of protons (nuclei in general) of compound-i, S i ( ) is the plateau value of the SRC curve-i and mm i is the molecular mass of compound i. Currently we offer the qsrc method for 2-compound mixtures (N=2) and 3-compound mixtures (N=3) Sample The Sample component is used to provide information about the sample and is only used for report purposes. See also H _2_

166 Time Domain Dynamics Method Center 2 Report [} 168] Data The Data component is used to load the SRC data of mixture and the two individual compounds. Figure 15.18: Data Dialog of the qsrc Method The input files are regular dps files (Data Point Storing) as also used in the TimeDomain1D method above. While in other time domain applications it is sometimes desired to discard some of the data points (e.g. measured at very small mixing times) one would not use this option here and always use all available data points. After successful loading the SRC curve of the mixture is shown on screen Analysis The Analysis component is used to model the SRC curve of the mixture as linear combinations of the SRC curves of the individual compounds. As described above a proper data scaling is needed. 166 H _2_002

167 Time Domain Dynamics Method Center Figure 15.19: Analysis Dialog Window of the qsrc Method There function to be used is already selected and shown for viewing. The number of protons and molecular masses of each compound must be provided. The confidence level of the fit indicates the probability that the true solution is within the obtained values for C 1 and C 2 +/- the calculated error margins View The data modelling is usually done within moments and the result can be viewed by the View command. The reconstructed SRC data of the mixture obtained from the data modeling can be compared to the input data. H _2_

168 Time Domain Dynamics Method Center Figure 15.20: Typical Result View of the qsrc Method Most important are the obtained quantification values, in this example c 1 and c which are rescaled such that c 1 +c 2 = 1. Details can be obtained by right mouse clicking to the modelled curve and selecting properties from the popup menu and via Report and Export Report A PDF report that contains loaded information, as well as results, can be generated in PDF format. Graphical components include the experimental SRC curve of the mixture and the one reconstructed from the data modelling. The user may select desired components in a dialog window. The name of the PDF file must also be specified. At the end of the report generation AcroRead is automatically launched to display the report. Some versions of AcroRead do not display the PDF file if it was not specified with its absolute path name, e.g. c:\tmp\test.pdf. If you don t have AcroRead on your computer an alternative PDF display program may defined under config/preferences/default PDF Viewer. The report (as well as export) contain some further interesting information such as the T 1 and plateau values of the individual SRC curves of the individual components. The reason is to ensure validity. 168 H _2_002

169 Time Domain Dynamics Method Center Figure 15.21: Extra Section in the Report of the qsrc Method Obviously the qsrc method would not work if the two components had the same T 1 values. It is required and indicated in the report that they differ at least such that the two values +/- their error margins do no overlap. Additionally the last few plateau values are checked for stability and quality, i.e. has the plateau really been reached or was the experiment time too short? Errors in the plateau directly translate into errors of the coefficients c 1 and c 2 and must be avoided. The relative standard deviations of each plateau should be as small as possible, which means that the last few plateau values are more or less identical Export Information, especially fit information and fit results, can be exported to a file on disk. The supported formats are text and xls (or xlsx). This allows users to present data with standard tools, e.g. chart diagrams in EXCEL. As already described in the previous section Report some extra information on differences of the T 1 values and plateau characteristics are included The qsrc(3) Method This method closely follows the description of the qsrc method, see previous sections. The only difference is that 3-compound mixtures instead of only 2-compound mixtures can be handled. In some cases the content c 3 of the third compound is actually known and shall not be modelled as such but kep constant while modelling c 1 and c 2. If c 3 shall be fixed to a known value it must be set in the dialog window. Note, that c1 + c2 + c3 = 1.0. H _2_

170 Time Domain Dynamics Method Center 170 H _2_002

171 Automation 16 Automation 16.1 Introduction The Dynamics Center is mainly used interactively, i.e. the user selects methods of interest and executes the individual components Sample, Data,... etc. Efficiency is increased if work is saved into a project, see Save or SaveAs from the popup menu if the right mouse button is clicked at a method. With Open such projects can be loaded again and executed. Often the same kind of analysis using the same parameters is wanted, so one only needs to change the names of the data sets under Data and keep everything else. The project files are stored in a property-value paired manner and are not intended for reading. With the DC 2.5. whenever a project is saved there is also an xml version saved under the same name but with the extension xml. These xml files can be sent into the Dynamics Center by drag & drop and would be executed just like a project that gets loaded and executed. As described in the next section one may however not just drop the xml but the folder in which the xml is stored. If this folder contains one or more further datasets then the xml is applied to them automatically. What can be done via drag and drop can also be done from a command line or script. This may be interesting for users who want to embed the Dynamics Center in their own environment, e.g. into a set of other scripts or programs that intend the Dynamics Center to execute a method up to Export and then access the exported data and do something themselves. An invocation of the Dynamics Center via script or other program would start a new instance of it, get something done and then keep it alive or shut it down. For more professional automation this is insufficient because one often needs to access functions of an already running program. In fact the Dynamics Center is implemented as a server which may be accessed from one or more clients. This can either be done with the help of a special script which is part of the installation or via a so-called socket communication. This is intended for programmers and is for example used by the Bruker IconNMR and biotop software Drag & Drop For illustration we assume that we are working on a Windows computer. Step 1: Get a suitable xml As an example we work in the Time Domain part and apply the TimeDomain1D method. When done we can use a right mouse button click to TimeDomain1D on the method tree and select SaveAs from the popup menu to save all work into a project file named c:\projects \TD_Latex.project. This will happen and in addition a corresponding xml files is also created: c:\projects\td_latex.project.xml Step 2: Dropping the xml file For simplification close and start the Dynamics Center again. In the windows file explorer navigate to c:\projects, locate c:\projects\td_latex.project.xml and drag & drop it on the TimeDomain1D method on the method tree. It will be executed right away as depicted in the following slide. H _2_

172 Automation Figure 16.1: Drop a XML File to a Method Naturally, one would drop the XML to a method from which it originated. When dropping to another method that would not fit an error message or warning is presented. It is allowed to drop on a method which is currently opened and in use you are then asked if it may be closed otherwise the drop action will not be finished. Step3: Apply the XML to a set of data all stored in a folder or its subfolders. The exercise in step 2 is somewhat boring because we only get what we have done before. We can however copy the xml to a folder which contains one or more other datasets. An example would be to copy the xml file originating from a T1 analysis of NMR data (say c: \projects\t1.project.xml) into a folder that contains a whole set of other T1 data (say e:\data \dynamic\nmr(2)\jochen_t1). We would then drag & drop the folder e:\data\dynamic \nmr(2)\jochen_t1 to the T1 method on the method tree. The procedure is depicted on the following slide. The folder e:\data\dynamic \nmr(2)\jochen_t1 contains 8 subfolders named 10, 1,..17 and the xml file named T1.project.xml. After dropping the folder e:\data\dynamic\nmr(2)\jochen_t1 to the T1 method in the general dynamics area T1.project.xml is now executed 8 times, i.e. the analysis is applied to all data found. 172 H _2_002

173 Automation Figure 16.2: Dropping a Folder Containing a XML file and many Subfolders As can be seen the T1 method is automatically added to the method tree and all 8 T1 methods are active. Since they have all been executed up to Export also 8 pdf and xls files have been generated. This is for example interesting if a large number of experiments und slightly different conditions have been measured and shall be compared. A few explanations need to be given: 1. The Dynamics Center looks for an xml, takes it as a template, applies it to all found datasets, each time replacing dataset name and name of report and export files. 2. The procedure fails if the xml and the type of data do not fit to each other. 3. The procedure fails if the xml is not unique. Several other standard xml files (peak lists etc.) are automatically filtered out and do not disturb. 4. To avoid that report or export files given in the xml overwrite each other, suitable new names are constructed by combining the report name with the dataset name. Example 1: If the report file name in xml is c:\tmp\alloutput\test.pdf and the dataset name is e:\data\sample\1\pdata\1\2rr the new report name is c:\tmp\alloutput \sample_1_1_test.pdf Since all NMR spectra have the same name (1r, 2rr, ) another name is constructed from the folder name above the expno folder and the expno and procno numbers. Example 2: If the report file name in xml is c:\tmp\alloutput\test.pdf and the dataset name is e:\data\td\latex.dps then new report name is c:\tmp\alloutput\latex_test.pdf Limitation Dropping a folder containing an xml file and several data file to a method assumes that the method can be executed to each of the datasets one by one. Some methods, e.g. qsrc and qsrc(3) need several datasets. In this case the method cannot be applied to each of the datasets at all. H _2_

174 Automation 16.3 Command line The descriptions given for drag & drop in the previous section can also be achieved in a shell window, e.g. command shell on Windows 7 which can be obtained by activating the Start button (lower left on screen) and entering cmd in the edit field. A black window (command shell, DOS shell) appears and commands can be entered. Suppose the Dynamics Center is installed in c:\bruker\dynamicscenter2.5, then in this folder there is a startup script named rundynamics.cmd. Here are some examples of command line applications 1. rundynamics c:\projects\td_latex.project.xml Starts another Dynamics Center and executes the specified xml file. It has the same effect as if this xml file had been dropped to the TimeDomain1D method. 2. rundynamics e:\data\dynamic\nmr(2)\jochem_t1 Starts another Dynamics Center which looks into the specified folder and if it finds a xml file it takes it and applies it ro each of the datasets. It has the same effect as if folder had been dropped to the T1 method. 3. rundynamics c:\projects\diffusion.xml is like the first example but rundynamics c:\projects\diffusion.xml e:\data\dynamic\nmr\dcref500\6\pdata\1\2rr again takes the specified xml but now applies it to the dataset given as the second argument. Something that can be entered in the command line can also be part of another script which itself can be executed in a shell or from another software. These techniques allow you for example to execute a method in a loop and compare the output. This can be part of an installation or system test. An inconvenience is that each time a Dynamics Center is started and you would have to manually shut it down afterwards. However the top section of each xml file contains a tag named AUTO_EXIT. By default this tag has the value <AUTO_EXIT>EXIT_IN_AUTOMATION_NO</AUTO_EXIT>. If you change the value of the tag manually to <AUTO_EXIT>EXIT_IN_AUTOMATION_YES</AUTO_EXIT> then the Dynamics Center exits itself after it has performed the required actions. If the Dynamics Center is already running and one would like to use it as a server, i.e. it shall keep running and execute incoming requests, then the simple command line example above cannot be used. Instead the caller (client) has to communicate with the server through a socket port and it has to know to which port the server is listening. To hide this complexity you may use a further script named runclient. Example: execute rundynamics to get a running Dynamics Center (=server) execute runclient with arguments which sends the arguments to the server Example: runclient executeinterfacefilexml c:\projects\diffusion.xml The first argument says the server shall execute a xml file, the second argument is the name of the xml file Actions like can again be embedded in other scripts which can be executed or called from other software. Some of the available client commands are: executeinterfacefilexml nameoffilexml executeserialinterfacefilexml nameoffilexml nameoflistfile add replace addmethodtotree nameofmethod removemethodfromtree nameofmethod 174 H _2_002

175 Automation closemethod nameofmethod exitdynamicscenter At Bruker we use the client server implementation of the Dynamics Center and the xml capabilities in applications such as InsightMR (reaction monitoring including acquisition), biotope (automation and optimization of 3D experiments) and diff (setup, run and analyse diffusion experiments). H _2_

176 Automation 176 H _2_002

177 InsightMR 17 InsightMR 17.1 Introduction InsightMR is used for online monitoring of processes as a function of time, e.g. chemical reactions. It is based on the Dynamics Center, IconNMR and TopSpin. Figure 17.1: InsightMR with IconNMR Embedded in the Dynamics Center Visible on screen is the Dynamics Center with IconNMR embedded. After experiments have been set up and started in IconNMR the display can be switched to the regular Dynamics Center using the Windows Control option on the method tree. While acquisition is going on, individual 1D spectra are generated and directly loaded. If signal regions have been defined they are automatically integrated and profiles are shown. It is possible to switch back and forth to IconNMR and even change acquisition parameters such as number of scans. H _2_

178 InsightMR Figure 17.2: InsightMR with Spectra and Profile Display 17.2 Using the Dynamics Center When using the Windows Control option on the method tree to switch to the spectra display the regular Dynamics Center is available. It is configured for InsightMR, i.e. there is no protein dynamics, general dynamics or time domain dynamics and the method tree only offers the Kinetics method. While an acquisition is running the spectra come to display one by one, signal regions can be defined as usual, see popup menu on spectra display and profiles are calculated and shown automatically. It is not needed to execute Analysis, View by hand as this is done automatically. It is possible however to do regular interactive work as usual and try for example to model the profiles. Limitations may exist or result in warnings. Accidentally closing the Kinetics method while spectra are acquired into it, is an example. The available features and the way of interactive working is pretty much identical to the Dynamics Center. In a few rare cases smaller differences (e.g. extra options in a popup menu) exist but should be easy to understand. You can for example define acquisition stop conditions based on signal integrals. 178 H _2_002

179 Tutorials 18 Tutorials Step-by-step tutorials are available for Protein Dynamics, General Dynamics, Kinetics and Time Domain Dynamics. They contain an illustrated PDF document that should be worked through in practise. Spectra used for the exercises are included. The tutorials can be downloaded from the Bruker Website: From there login and select for example the PC Dynamics Center software. The tutorials are identical for different computer platforms Protein Dynamics Tutorial Experimental pseudo 3D spectra (NOE, T1 and T2) and other needed information (amino acid file, pdb file) are available. The sample used is ubiquitine. 2D-Ref 2D hsqc spectrum with a good peak list. 185-NOE Pseudo 3D data for NOE analysis T1 Pseudo 3D data for T1 analysis T2 Pseudo 3D data for T2 analysis. ubi_fasta.txt Fast a file of ubiquitine, contains amino acid sequence. 1UBQ.pdb Ubiquitine pdb file (no protons). 1UBQ_H.pdb Ubiquitine pdb file with protons. tutorial_pdc.pdf Illustrated step-by-step exercises General Dynamics Tutorial Experimental pseudo 2D spectra (T1, diffusion) are available. T1_decay Pseudo 2D spectrum for T1 analysis. Diffusion Pseudo 2D spectrum for diffusion analysis. tutorial_dc.pdf Illustrated step-by-step exercises. H _2_

180 Tutorials 18.3 Kinetics Tutorial Experimental series of 1D spectra are available. AA_25C_500mM 1D spectra of hydrolysis of acetic anhydride (CH 3 CO) 2 O + H 2 O 2 CH 3 CO 2 H (CH 3 CO) 2 O + H 2 O 2 CH 3 CO 2 H tutorial_kinetics.pdf Illustrated step-by-step exercises Time Domain Dynamics Tutorial An experimental decay curve and a step-by-step exercise are available. Sample A.dps tutorial_time_domain_dynamics.pdf 180 H _2_002

181 Further Information 19 Further Information Author Dr. Klaus Peter Neidig Web page Phone H _2_

182 Further Information 182 H _2_002

183 Contact 20 Contact Manufacturer Bruker BioSpin GmbH Silberstreifen 4 D Rheinstetten Germany WEEE DE NMR Hotlines Contact our NMR service centers. Bruker BioSpin NMR provides dedicated hotlines and service centers, so that our specialists can respond as quickly as possible to all your service requests, applications questions, software or technical needs. Please select the NMR service center or hotline you wish to contact from our list available at: Phone: nmr-support@bruker.com H _2_

184 Contact 184 H _2_002

185 List of Figures List of Figures Figure 1.1: A View of the Dynamics Center with Access to Protein and General Dynamics... 9 Figure 1.2: A View of the Method Tree, with the T2 Method Node Opened Figure 2.1: The Help Pull-down Menu Figure 3.1: Listing of Installed System Services Figure 4.1: Starting the Dynamics Center via Applications/Bruker TopSpin Figure 4.2: Starting the Dynamics Center via Desktop Icon Figure 4.3: TopSpin Flow Bar for Dynamics Center Support Figure 5.1: GUI Components of the Dynamics Center Figure 5.2: Setting the Spectrum Default Path Figure 5.3: Standard Data Slider Figure 5.4: Sliders Added to the Main Tool Bar Figure 5.5: Left and Right Click to the Method Tree Figure 5.6: First Tab on the Left: File System Tree Figure 5.7: File System Tree with Popup Menu Figure 5.8: Drag & Drop Target Positions Figure 5.9: Drag & Drop a Series of 1D Spectra to the Kinetics Method Figure 5.10: Options to Drop a Project File Figure 5.11: Linked Cross-hair Cursor Figure 5.12: The Icon Tool Bar Figure 5.13: Context Sensitive Popup Menus Figure 5.14: Dynamics Center Before Version Figure 5.15: Dynamics Center Figure 5.16: Available User Interface Types Figure 6.1: Color Codes indicate the State of a Method Figure 6.2: Data Dialog/Spectra Tab for the Selection of Spectra Figure 6.3: User Defined Peak Areas: Move/resize with Cursor Figure 6.4: Data Display with Slider Figure 6.5: The Popup Menu of 1D and pseudo 2D Spectra offers Stacked Plot Display Figure 6.6: Stacked Plot Display of a Series of 1D Spectra Figure 6.7: Example of an Analysis Dialog Window, T1 Method Figure 6.8: ILT Tab of the Analysis Dialog Window Figure 6.9: View Dialog Window to customize the Result Display Figure 6.10: Fit Display Objects also have Context sensitive Popup Menus Figure 6.11: Loading external Data, e.g. simulated REDOR Curves Figure 6.12: Results of a T1rho Analysis with 2D Plots from Fit (upper right) and ILT (lower left) Figure 6.13: Possible Pulse Sequence to measure Diffusion Constants Figure 6.14: Data Dialog of the Diffusion Method Figure 6.15: The 3D Display can be selected from the Popup Menu of a 2D Display Figure 6.16: Example of Pseudo 3D i-dosy Data Set with COSY Spectra in F2, F H _2_

186 List of Figures Figure 6.17: ILT Tab of the Analysis Dialog of the Diffusion Method Figure 6.18: ILT with Log Grid, (middle) Compared to Fit (left) Figure 6.19: As Previous Figure but with Small Alpha and Tikhonov Regularization Figure 6.20: Problematic ILT Result, Some Peaks Shifted and Back Calculation Deviates from Input Figure 6.21: DOSY Plot Tab of the View Dialog of the Diffusion Method Figure 6.22: Typical Diffusion Result View with DOSY Plots based on Fit and ILT Figure 6.23: 3D DOSY Display selected from the Popup Menu of the 2D Display Figure 6.24: Dynamics Center Interface in the Diff Software to Set Up and Launch a Complete Analysis Figure 6.25: Data Dialog of the Kinetics Method with the Option to load from Text File Figure 6.26: Simple Tool to create a Text File with Spectra Names and Time Values Figure 6.27: Peaks Tab of the Data Dialog of the Kinetic Method, Top Part Figure 6.28: Integrals Tab of the Data Dialog of the Kinetic Method, Top Part Figure 6.29: The Spectrum Popup Menu contains Tools to add Integration Areas Figure 6.30: Adding multiple Integration Areas using the Mouse Figure 6.31: Updated Display after having added Integration Areas Figure 6.32: Bottom Part of the Peaks Tab of the Data Dialog Window Figure 6.33: Shifting Signals of Acetic Anhydrite (left) and Acetic Acid (right) Figure 6.34: The Analysis Dialog Window of the Kinetics Method Figure 6.35: The Normalization Tab of the Analysis Dialog Window of the Kinetics Method Figure 6.36: Customizing the Fit Curve Display via the Units & Label Tab Figure 6.37: Example of using 2D Spectra for Reaction Monitoring Figure 6.38: Successful CP Curve Fit of Alanine Resonances Figure 6.39: REDOR Data Dialog Window Figure 6.40: REDOR Analysis Dialog Window Figure 6.41: Typical REDOR Result. The Initial Points are Fit to the Second Moment Figure 6.42: Simulated REDOR Curves loaded from a csv File Figure 6.43: Export Dialog Window with an Interface to Simpson Figure 6.44: The Arrhenius Data dialog Figure 6.45: The Arrhenius Analysis Dialog with a tab named Arrhenius Analysis Figure 6.46: Arrhenius Plot of one Peak Figure 7.1: Color Codes Indicate the State of a Method Figure 7.2: Tab Oriented Dialog Window to Describe the Sample Figure 7.3: Data Dialog for the Selection of Spectra Figure 7.4: Comparison of a T1 Fit Based on Intensities (A), Area Integrals (B), Shape Integrals (C) and De-convolution (D), Differences of T1 are Less Than 1% Figure 7.5: Data Display with Slider Figure 7.6: Example of the Analysis Dialog Window, here T2 Relaxation Figure 7.7: Example of the User Defined Function to Analyze J-modulated 1H-15N HSQC Spectra Figure 7.8: Results of fitting a User Defined Function to J-modulated 1H-15N HSQC Peaks Figure 7.9: View Dialog Window to Customize the Result Display Figure 7.10: Typical Result Display with Spectrum, Histogram and Fitted Curve H _2_002

187 List of Figures Figure 7.11: Fit Display Objects also have Context Sensitive Popup Menus Figure 8.1: The Combined NOE/T1/T2 Method is Part of the Method Tree Figure 8.2: Data Dialog Window to Select Project Files to be Used for NOE/T1/T2 Modelling Figure 8.3: Successful Loading of a Group of NOE, T1 and T2 Projects Figure 8.4: The Settings Tab is Used to Define Some Global Parameters Figure 8.5: The TauC Tab is Used to Select Proper Residues for the Global TauC Calculation Figure 8.6: The Reduced SD Tab is related to Spectral Density Calculations Figure 8.7: View Dialog (Rex tab opened) to Define Details of the Result Display Figure 8.8: Example of a Result Display Showing Various Types of Objects Figure 8.9: Example of a Result Display Illustrating the Combined Interpretation Figure 8.10: A Simple Structure Viewer Based on Jmol Figure 8.11: Partial View of a PDF Report Page Figure 9.1: The Side Chain Dynamics Method is Part of the Method Tree Figure 9.2: Data Dialog Window to Select Project Files to be Used for Sidechain Dynamics Modeling Figure 9.3: Successful loading of two Groups of T1 and T1rho Projects Figure 9.4: Warning if Sequence Numbers contained in Peak Annotations are not unique Figure 9.5: Example of a Histogram Display constructed from sorted Peak Annotations Figure 9.6: Analysis Dialog Window of the Sidechain Dynamics Method Figure 9.7: Histogram Display of Order Parameters and internal Correlation Times Figure 9.8: Part of a typical Sidechain Dynamics Report Figure 10.1: Example of a Threshold based Peak Picking of a NOE Spectrum Figure 10.2: Analysis Dialog Window of the NOE build-up Method Figure 10.3: View Results after fitting NOE build-up Curves Figure 11.1: Analysis Dialog Window of the SN Comparison Method Figure 11.2: View Results after S/N Calculation as a Function of D Figure 12.1: CEST is Located in the Protein Dynamics Method Tree Figure 12.2: Dialog Window for CEST Analysis, Systematic Variation of Start Values for we Activated Figure 12.3: The Start Parameters Tab Figure 12.4: Successful Fit of the CEST Profile Closest to the Cursor Figure 13.1: The Multi T2/Kd Data Dialog Figure 13.2: The Multi T2/Kd Analysis Dialog Contains a Tab Named Kd Anlaysis Figure 13.3: Result View of the Multi T2/Kd Method Figure 14.1: Standard Export dialog Window of any Relaxation Method Figure 14.2: NESSY Dialog Window to Import Output from the Dynamics Center Figure 15.1: Example of a Header Section in a dps File Figure 15.2: Figure 15.3: Example of a dps File Containing Time Values in Column 1 and Amplitudes in Column Example of a dps File Containing Time Values in Column 2 and Amplitudes in Column Figure 15.4: The Analysis Dialog Window Used for Time Domain 1D data Figure 15.5: Dialog Window to Setup an ILT Calculation for Time Domain Data Figure 15.6: The ILT Regions Tab is Part of the Analysis Dialog Window H _2_

188 List of Figures Figure 15.7: View Dialog Figure 15.8: Typical View after Time Domain Analysis Figure 15.9: Typical Display of analysis results of Time Domain Dynamics Figure 15.10: Data Dialog of the TimeDomain2D Method Figure 15.11: The Dimensions/Kernels Tab of the Data Dialog Figure 15.12: The Analysis Dialog Window Used for Time Domain 2D data Figure 15.13: ILT Regions Tab to Define 2D Regions Figure 15.14: The View Dialog Window Figure 15.15: Typical Result Display of a FISTA Calculation Figure 15.16: Interactive Change of Regions Causes Change on Tree Figure 15.17: Principle of the Data Modeling of a 2-Component System Figure 15.18: Data Dialog of the qsrc Method Figure 15.19: Analysis Dialog Window of the qsrc Method Figure 15.20: Typical Result View of the qsrc Method Figure 15.21: Extra Section in the Report of the qsrc Method Figure 16.1: Drop a XML File to a Method Figure 16.2: Dropping a Folder Containing a XML file and many Subfolders Figure 17.1: InsightMR with IconNMR Embedded in the Dynamics Center Figure 17.2: InsightMR with Spectra and Profile Display H _2_002

189 Index Index Symbols.dps file , 166.sig file /etc/sysinfo Numerics 2D ILT D plot D display A AA-Sequence About Abragam-Weibull-Exponential abundant acetic acid Acrobat reader AcroRead... 54, 100 adaptive peak picking... 40, 89 add peaks Adobe Reader AIC... 47, 71, 106 alpha... 48, 59, 153 amino acid sequence Analysis... 21, 45, 93, 148, 160, 166 Angstrom , 134 anhydrite Anisotropic modelling... 88, 112 Annotate ASAP automated picking AV NEO B B value... 55, 61 B values Bezier-smoothing... 30, 32 B-factor big delta biotop BMRB C CEST chi-squared... 48, 95 client cmd shell CodeMeter compensated pulse sequence confidence interval... 48, 95 confidence level... 48, 95, 150 CONST constant duration Context sensitive CONTIN COSY i-dosy cpmg... 39, 43, 91 Cross Polarization cross-hair cursor Cross-link density cross-relaxation csv... 52, 80 D D Data... 21, 40, 89, 146, 158, 166 data-slider... 44, 92 decay rates derivatives diff software... 56, 64 Diffusion Diffusion manual diffusion tensor DiffusionPar dilute spins distance dmg Dosy plot DOSY plots double resonance dps , 147, 158 Drag & Drop... 25, 26 drag and drop E eigenvalue environment variable Error bars... 50, 96 error estimation... 48, 95 EXCEL Explorer expno Exponential Exponential decay Export... 21, 54, 100, 158, 163, 169 F F FASTA file system tree Finder first first order kinetics H _2_

190 Index FISTAFISTA Fit Functions , 150 Flexlm FLEXlm Diagnostics Floating licenses Flow bar Fourier... 9 FQ3LIST G Gaussian... 43, 58, 90, 151 General Dynamics... 9 global correlation time goodness of fit... 53, 99, 157 gradient strength grid... 48, 160 grid type... 48, 58, 153 H Help Help/About histogram hostid hsqc_cest_etf3gpsitc3d I icon tool bar IconNMR , 177 i-dosy ILT... 57, 149, 151, 153 InsightMR... 10, 177 interleaved mode intermediates inverse Laplace Transform... 57, 149, 153 Inversion recovery , 151 ipconfig isotopomers Isotropic modelling J Jmol jre K kernel kernel compression kernel matrix kinetics L Levenberg-Marquardt License Agreement license.dat line shape analysis linear Linux CentOS Lipari-Szabo mapping Lists tab... 43, 91 little delta... 54, 61 LM_LICENSE_FILE lmutil... 15, 16 local correlation time log grid logarithmic loop duration... 43, 91 Lorentzian... 43, 90, 151 M Mac MAC address main menu bar Manual manual peak analysis Marquardt MAS matching measure distance... 43, 90 Method Center method oriented... 9 method tree... 10, 19 minispec... 9, 147 model-free Molecules... 23, 54 Monte Carlo... 48, 95 N NESSY NNLS Node locked licenses NOE buildup non-linear fit... 48, 95 number of mixing times... 43, 91 number of nuclei P Partial inversion recovery pdb file PDF report... 54, 100, 158, 163, 168 peak areas peak deconvolution... 43, 90 peak integrals... 40, 90 peak intensities... 42, 89 peak popup peak snapping... 41, 89 peak tracking... 42, 67, 70 peaklist.xml Peaks H _2_002

191 Index Peaks tab Perspective PFG Physical Address Positional Variability preferences... 19, 20 PreView project project file Protein Dynamics Python... 47, 93 Q q values qsrc quadrupolar relaxation R R R random generator RasMol/Chime reaction monitoring recycle delay REDOR Reduce reduced spectral densities registration regularization... 48, 58, 153, 160 regularization parameter relax repetition experiments... 43, 91 Report... 21, 54, 63, 100, 158, 163, 168 residue number... 32, 99 Resume runclient rundynamics S Sample... 21, 40, 88, 129, 133, 146, 158, 165 Saturation recovery , 151 Scaling second moment second order kinetics SEQ server Settings shape integration Shapiro Wilk... 50, 156 Sidechain Dynamics sig , 158 Simplex Simpson singular value decomposition slider... 20, 44, 92 smoothness SOFAST software CD Spectra tab spectral density Spectrum Default Path SRC start-up script Student-s-T... 48, 95, 150 Suspend SVD T t1par... 56, 64 tab separation TD TD effective Thikonov threshold based picking Tikhonov Time Domain Dynamics... 9, 145 time stamp toggle TopSpin... 17, 20, 36, 56 TopSpinDC.xml Tutorials U uncertainties... 43, 48, 95 user defined V variance averaging vclist... 43, 91 vdlist View... 21, 50, 96, 132, 134, 155, 161, 167 visibility W Windows Explorer X XEASY xls... 54, 100, 158, 163, 169 xlsx... 54, 158, 163, 169 xml xml files xml interface Z Zoom H _2_

192 Index 192 H _2_002

193 H _2_

194 Bruker Corporation Order No: H