ADVANCED DATA PROCESSING

Transcription

1 ADVANCED DATA PROCESSING Making Backups The NMR lab periodically backs up data on CD-ROMs for your convenience ([Files] [Oldfids]). If you would like your own personal backups, you can do this via the computer called spectrum. 1. Insert a PC-formatted 100MB zip disk into the zip drive. 2. Select [Unix Shell]. 3. In the shell window, change to the directory from which you want to copy, e.g.: > cd /data/6-lettercode (for current data) > cd /v3data/6-lettercode (VAC-300 files) > cd /v2data/6-lettercode (VAC-200 files) > cd /old/b8/6-lettercode (for archived data) > cd /cdrom/cdrom0/6-letter code (for data on the CD-ROM drive) 4a. Copy files to the zip drive: > cp -r filename /rmdisk/zip0 (for a single file) > cp -r * /rmdisk/zip0 (for all files) 4b. For data archived on the CD-ROMs, you will have to type: > tar cf - filename (cd /rmdisk/zip0; tar xf - ) 5. To eject your zip disk, type: > eject zip0 Printing Insets 1. Plot the full spectrum manually: pl pscale pirn ppa (Do NOT type page!) 2. Put the cursors around the region you would like printed as an inset and type inset. 3. Use the mouse buttons to move the spectrum to the area where you would like the inset to be printed. The left mouse button moves the inset region back and forth on the screen. The middle mouse button adjusts intensity. The right mouse button adjusts the width of the spectrum. To adjust the vertical position of the spectrum, use the vp command. (You can also do all of this manually by adjusting the following values: sc, wc, vp) 4. Plot the inset and send it all to the plotter: pl pscale pirn page 5. To get back to the original spectrum, type vp=12 f full 1

2 Arrays: Creating, Processing, Displaying, and Plotting CREATING ARRAYS Several spectra can be combined into one dataset of arrayed spectra. This is very useful for comparing spectra collected under the same conditions, e.g., to compare peak heights over a series of spectra. 1. Load each FID into a separate experiment, leaving exp5 empty. Do NOT process your data. (If you do not have enough experiments, create them by typing cexp(#) where # is the experiment number you want to add.) 2. Join experiment one: jexp1 and type clradd. This deletes exp5. 3. Type add( new ) in experiment 1. Join all the other experiments: jexp2 and type add( new ) etc. until all fids have been added to experiment Join experiment five: jexp5 5. Type setvalue( arraydim,#, processed ) where # is the total number of fids that you have added. PROCESSING AND DISPLAYING ARRAYED DATA The data in exp5 will now behave as an arrayed experiment. Save it if you wish. wft will process all the fids wft(1) will process just the first fid, wft(2) will process the second, etc. ds(1) will display the first fid, ds(2) will display the second, etc dssa will display all fids stacked vertically dssh will display fids in a line horizontally PLOTTING ARRAYED DATA To plot the entire array as displayed on the screen, type: pl( all ) pscale page To plot individual spectra, display the spectrum with the ds(#) command, then pl pscale page Comparison of Spectra METHOD 1 If you would like to compare the same regions of two spectra and want the vertical scales and ppm ranges to be identical, you can create saved regions. This is especially useful in viewing 2D datasets where you routinely want to look at a particular region of the spectrum. 1. Call up each spectrum in different experiments (eg. exp 1 and 2) and process them as desired. 2. In experiment 1, type s1 to save the region (that s a one, not an L). You can save up to 9 regions. The s command saves a copy of all the current display parameters. The saved parameters are dataindependent, i.e., the fid is not being saved. 3. Type md(1,2) to move those display parameters (all the s files you created) from experiment 1 into experiment Go to experiment 2 (jexp2) and type fr1 or r1 to recall the data region. fr1 does a full recall. r1 will not recall phase, drift correction, integrals, or reference parameters. It does recall chart info (wc, sc, wp, sp, etc), horizontal and vertical offsets, vertical scaling, and ai/nm mode. 5. Type ds (or dconi if 2D) to display the region. 2

3 METHOD 2 In each spectrum, set sp and wp to the desired chemical shifts, e.g., sp=0p wp=10p. sp is the start of the chart in ppm wp is the width of the chart in ppm PLOTTING SPECTRA AS STACKED PLOTS ON THE SAME PAGE For a set of three spectra in exps 1-3, after each has been set to the same regions using the methods above, join each experiment and plot the spectrum without typing page until the final spectrum. Ex: For the bottom spectrum, jexp1 type ds and adjust the height with the middle mouse button bearing in mind that 3 spectra need to fit on the same page pl pscale For the middle spectrum, jexp2 type ds and adjust the height with the middle mouse button vp=70 pl For the top spectrum, jexp3 type ds and adjust the height with the middle mouse button vp=140 pl page Experiment with vp (vertical position) and vs (vertical scale) to adjust the spacing between spectra. For 4 spectra, try vp s of 45, 90, and

4 Creating Postscript Files Spectra, both 1D and 2D, can be saved as postscript files. This is useful when you need to make figures for a paper and would like to edit the file with labels, colors, etc. 1. Log in to any workstation labelled fid in the NMR lab. 2. Choose a PS-AR plotter (for landscape plots) using the [Change Plotter] button. 3. Process your spectrum. Display it the way you want it for the figure. 4. Make sure you are in your data directory. [Main Menu] - [File] - [Data] or by typing: cd( /data/6-lettercode ) 5. Type in all the commands necessary for plotting the spectrum, but give page a filename, e.g.: For 1D spectra: pl pscale pirn ppa ppf( top ) page( filename.ps ) For 2D spectra: pcon(10,1.2) pap page( filename.ps ) 6. Transfer your files to a zip disk. See the section on Making Backups. Creating ASCII Files In some cases, you may want to save your spectra as an ASCII file. This is a file consisting of 2 columns, the first is the ppm value and the second is the intensity. 1. Log in to any workstation labelled fid in the NMR lab. 2. Recall your FID. 3. Change the zerofill number (fn) to something more reasonable, e.g. fn= This method creates a new entry for each datapoint; this is the reason for reducing the fn number. 4. Reprocess (wft) and make sure you do not lose too much resolution. 5. Make sure you are in your data directory. [Main Menu] - [File] - [Data] or by typing: cd( /data/6-lettercode ) 6. Type writespec( filename ) in the command line. Use a different filename than that of your original fid. Otherwise you will not be able to read the original fid later on. 7. Transfer your files to a zip disk. See the section on Making Backups. 4

5 Solvent Subtraction Filtering Sometimes it is necessary to remove a large solvent peak that was not suppressed sufficiently during acquisition. This can be done post-acquisition using digital filtering. (p. 238 of System Operation manual) There are two types of algorithms for digital filtering: lfs (low-frequency suppression) in which a low-pass filter is applied to the FID. This filter attenuates all signals that lie outside the filter, leaving only the on-resonance solvent signal. This filtered FID is then subtracted from the original FID to remove the solvent peak contribution. zfs (zero-frequency suppression) in which the FID is also low-pass filtered, but then fitted with a polynomial (specified by the parameter ssorder). This is then subtracted from the original FID. This gives a FID in which the signal that is exactly on-resonance has been removed. (Use this when you have moved the tof to the solvent signal.) 1. Recall the fid and process it. 2. Type addpar( ss ) This creates the following parameters: ssfilter specifies the full bandwidth of the low-pass filter; default is 100 Hz. sslsfrq specifies the location of the center of the solvent-suppressed region. Setting this to a non-zero value shifts the solvent-suppressed region by sslsfrq Hz. ssntaps specifies the number of taps (coefficients for the digital filter). Default is 121, but the value can range from 1 to np/4. ssorder specifies the order of the polynomial used to fit the FID 3. Decide which algorithm (lfs or zfs) is appropriate for your dataset and enter the correct parameters according to the following guidelines: For zfs, ssfilter is set to a value other than not used. Default is 100 Hz. sslsfrq= n (suppresses a region centered around the transmitter frequency (tof). ssntaps--use values between 3 and 21 ssorder-- set to a value from 1 to 20. For lfs, ssfilter is set to a value other than not used ssorder= n To set sslsfrq: Set axis= h Put the cursor in the middle of the spectrum (i.e., find the tof) by typing: cr=sw/2-rfl+rfp Put the right cursor on top of the peak to be subtracted. (If the peak is to the left of the tof cursor, use the left cursor on the peak to be subtracted and the right cursor on the tof.) Repeat this step if you have a second peak you would like to subtract. Set sslsfrq=delta (Use -delta if peak is to the left of tof; use an array if you are subtracting 2 peaks. ssntaps--default value is suitable 4. Type wft to perform the subtraction. 5

6 Spin Simulation You can similate a spectrum of up to EIGHT closely coupled, non-equivalent spins (ABCDEFGH) using the vnmr software (p. 359). Equivalent spins can be treated by magnetic equivalence factoring (e.g., A3B2CD3). Nuclei are treated as different types if there is at least one spare letter in the alphabet between their groups (e.g. ABD and ABX). Frequencies, intensities, energy levels, and transitions can be listed and simulated spectra can be displayed and plotted. Parameters can be iteratively adjusted to approach an experimental spectrum. Example of an AX2Y system using menu buttons: 1. Load an FID and transform it. 2. Expand the multiplet you wish to simulate. 3. Adjust the threshold so that all peaks are picked. 4. Enter dll to display a line listing (will be used later). 5. Select [Main Menu] - [Analyze] - [Simulation] You are now in the Spin Simulation Main Menu. 6. Select [Spin System] - choose the spin system (use [other] to get more options) The example used here is an AX2Y system. 7. Select [Set Params]. The spectrum will reappear. 8. Click the left button on the center of the multiplet and enter A=cr This sets the chemical shift of spin A to the position of the cursor 9. Click the left button on the center of the left-most line, and the right button in the center of the second left-most line, enter JAY=delta This sets the coupling constant to match the difference frequency. 10. Click the right button on the center of the third line and enter JAX=delta. 11. Select [Main Menu] - [Analyze] - [Simulation] - [Show Params] - [Simulate] The simulated spectrum will appear. 12. To continue iteratively, type interate? The status window will display: iterate= A, JAX, JAY 13. Select [assign] - [auto assign] The assign macro is executed, assigning the lines from the dll listing to the lines from the previous simulation. 14. Select [iterate] 15. Select [list] This listing contains the values of the A, JAX, and JAY parameters that give the best iterated fit to the experimental spectrum. 6

7 If you want to simulate a spin system from scratch, load a H1 dataset at the same field strength for which you would like to simulate. This will ensure that your coupling constants are correct. 1. Click on [Main Menu] - [Analyze] - [Simulation] You are now in the Spin Simulation Main Menu. 2. Click on [Spin System] - choose the spin system (use [other] to get more options) 3. Click on [Set Params] The spectrum will reappear. 4. Enter chemical shifts for each peaks in Hertz: A=#### B=#### X=#### etc. 5. Enter the coupling constants in Hertz. JAB=#### JAX=#### JBX=#### etc. 6. Click on [Main Menu] - [Analyze] - [Simulation] - [Show Params] - [Simulate] The simulated spectrum will appear. Deconvolution Complex patterns of multiplets can be deconvoluted into their component parts for integration purposes. The following parameters are available for each line: Frequency (in Hz) Intensity Linewidth (in Hz) at half-height of line Gaussian fraction of line (0.0 for completely Lorentzian line; 1.0 for completely Gaussian line) All parameters can be fit simultaneously or selected parameters can be removed from the fit. A linear baseline correction is always added to the fit to avoid errors from baseline effects. The following text files are used: fitspec.inpar--contains starting parameters fitspec.indata--contains point-by-point intensity of the spectrum fitspec.outpar--contains final parameters after a fit mark1d.out--contains results of a mark operation For best results, use fn 2*np For complex problems, use usemark to set the best possible guesses (see below). Example using the menu buttons: 1. Load an FID. 2. Enter fn= Enter lb= Transform the fid: wft 5. Expand on the region of interest so that it fills the screen. 6. Type ai for absolute intensity mode. 7. Choose a threshold value that picks all the peaks of interest and type dpf to check this. 7

8 If all peaks of interest are not picked, you will need to use the [Mark] function. Put the cursor on each peak you are interested in and select [Mark]. This will create a file called mark1d.out in your vnmrsys/exp# directory. For each deconvolution you plan to do, this file must be zeroed such that there are no entries in it as the [Mark] command appends to the file: mark( reset ) 8. Click on the buttons: [Main Menu] - [Analyze] - [Deconvolution] 9. Click on [Use Line List] or [Use Mark] depending on the outcome of step 7. This creates a line list and file containing the starting point for the deconvolution and automatically measures the linewidth of the tallest line on the screen and uses that as the starting linewidth for the calculation. 10. Click on [Fit] This performs the deconvolution. When finished, the calculated spectrum is displayed and the numerical results appear in the text window. 11. To plot the results, click on [Plot] This plots the original spectrum, the calculated spectrum, and each of the component lines, along with the numerical results. At the end of this operation, the original spectrum is again displayed. 12. To return to the calculated spectrum, click on [Main Menu] - [Analyze] - [Deconvolution] - [Show Fit] 13. To view the original and calculated spectra simultaneously, click on [Add/Sub] 14. Select [sub] from the menu to view the difference between the two. More details: slw--starting linewidth. Set automatically using the menu buttons or can be bypassed using the usemark macro with 2 cursor input. Typical value is 1. fitspec performs deconvolution by fitting experimental data to Lorentzian and/or Gaussian lineshapes. This macro uses the file fitspec.inpar (containing information on frequency, intensity, linewidth, Gaussian fraction) which must be created before the calculation. Any number followed by an asterick(*) is held fixed during the calculation. fitspec creates the file fitspec.indata, a text representation of the spectral data. After the calculation, the results are contained in the file fitspec.outpar. The setgauss macro modifies the output of the last deconvolution (fitspec.outpar) and makes it the input for subsequent analysis (fitspec.inpar), after first modifying the Gaussian fraction: To allow this fraction to vary, use the format setgauss(fraction) where fraction is the Gaussian fraction of the lineshape, from 0 to 1, e.g., setgauss(0.4) To fix the fraction, use the format setgauss( fraction*), e.g., setgauss( 1.0* ) Display and Plotting Deconvolution: The results are written into the file fitspec.outpar. The macro showfit converts the data into a useful format for analysis and printing. The command dsp( fitspec.outpar ) displays the theoretical spectrum. The macro plfit produces a complete output plot with the observed spectrum, the full calculated spectrum, each individual component, as well as the numerical results of the analysis (or use the [Plot] button in the Deconvolution menu). 8

9 A Customization of VNMR which Facilitates the Structure Elucidation of Organic Compounds --written by Ion Ghiviriga, Dept of Chemistry, Univ. of Florida Type "custom" on the command line to activate this program. Type "rm_custom" when you are finished. This is available ONLY on the offline workstation fid. What It Does The program does what a spectroscopist would do with a pencil and a ruler on top of the spectra, but nicer and faster. It facilitates the structure elucidation of organic compounds, where several 1D and 2D spectra are used. First, the chemical shifts and the splitting patterns are marked by clicking on the spectrum and in the menus. For 2D spectra, chemical shifts in both f1 and f2 can be marked in this way. The data can then be saved, displayed/plotted with the spectrum, or written in an ascii file which can be imported in any text editor. All the operations are menu-driven. Although the usage is self-explanatory, help files are provided. Type "help" to access them. Files The files used by the program must be in the same directory as the data. Create a directory for each sample. All the editing of the chemical shifts and couplings can then be archived with the data. Files used for displaying chemical shifts/couplings: cs_js chemical shifts and couplings displayed with the 1D spectrum. Created by marking on the spectrum, or peak-picking. Copied to cs_[tn/dn] with Save. cs1/cs2 chemical shifts in f1/f2 displayed with the 2D spectrum. Created by marking on the spectrum, or by loading from a cs_[tn/dn] file. Copied to cs_[tn/dn] with Save. Files used for storing chemical shifts/couplings: cs_[tn/dn] chemical shifts saved for the nucleus [tn/dn] (default name). Files used for exporting chemical shifts/couplings to text editors: chemshifts_[tn/dn] a header followed by a list of chemical shifts couplings_[tn/dn] a header followed by a list of chemical shifts and couplings. 9

10 Typical Usage PROTON SPECTRA 1. Move to the directory containing the spectra for a compound, (e.g. h.fid, c.fid, dqcosy.fid, ghmqc.fid). Remember that the program reads from and writes to the current directory, so you have to be in the directory for a compound when using it. 2. Load h.fid into exp1 and wft with a window function which enhances the resolution. 3. Click [Main Menu] - [Display] - [Edit Chemical shifts] - [Mark] 4. Mark the chemical shifts and couplings on the proton spectrum: Place two cursors around a multiplet and click [Expand]. Put one cursor in the center of the multiplet and click [Mark shift]. Define the multiplicity, e.g. for a doublet of heptets: put two cursors on two of the lines which define the doublet then click doublet. put two cursors on the OUTER lines of the heptet and click other. Answer hp for the symbol and 7 for the number of lines. Click [DONE]. If the pattern displayed on top of the peaks matches the lines, drag to Yes in the window which pops up. If you drag to No (or Maybe :)) no marking will be retained and you can repeat (4). Repeat (4) until all the peaks are defined. If there are any mistakes, at any time you can unmark the peaks between the cursors by clicking [Unmark]. 5. Save what you have marked. Click [Return] - [Save]. Type <Enter> to go for the default name, cs_h1. 6. The coupling pattern will be displayed with the spectrum. In the Interactive Display Menu there is a button to turn the display of the coupling pattern on/off. 7. Click [Main Menu] - [Display] - [Plot] - [CS+JS] - [Scale] - [Page] to plot the spectrum with the coupling pattern on top of the peaks. 8. Click [Main Menu] - [Display] - [Edit Chemical shifts] - [Write chemical shifts and couplings]. Type <Enter> for the default name, couplings_h1. FTP this file to the computer on which you do the text editing and import it into your report as ascii. CARBON SPECTRA 9. Load c.fid into exp2 and wft. 10. Place a threshold at a level which selects the peaks. 11. Click [Main Menu] - [Display] - [Edit Chemical shifts] - [Get] - [Peaks] higher than th. 12. Use the [Mark Menu] to unmark the solvent peaks. 13. Save/display/plot/write the peak editing as for the proton spectra (5-8). 10

11 COSY SPECTRA 14. Load a homocorrelation spectrum in exp3 and transform. 15. Click [Main Menu] - [Display] - [Size] - [Homonuclear] to select a size which allows room for displaying the chemical shifts. 16. Click [Main Menu] - [Display] - [Edit CS] - [Get] - [From File - cs_h1] - [Apply]. 17. The coupling pattern will be displayed with the spectrum. In the Interactive 2D Display Menu there is a button to turn the display of the chemical shifts on/off. 18. In the Plot 2D Menu there is a button [CS] to plot the chemical shifts with the spectrum. HETERONUCLEAR SPECTRA 19. Load a heterocorrelation experiment, let's say hmqc in exp4 and transform. 20. Click [Main Menu] - [Display] - [Size] - [Heteronuclear] to select a size which allows room for displaying the chemical shifts. 21. If there is a carbon spectrum available, click [Main Menu] - [Display] - [Edit CS] - [Get] - [From File - cs_c13] - [Apply to f1]. 22. If there is no carbon spectrum available, cs1 contains now proton chemical shifts and has to be reset. Click [Main Menu] - [Display] - [Edit CS] - [Get] - [Reset cs in f1]. 23. Mark the chemical shifts in f1. In the [Edit Chemical Shifts Menu], click [Mark & Save] - [Mark CS in f1]. Use Full/Expand to get a suitable display of the spectrum. Put a cursor on the peak and click [Mark shift]. Repeat until all peaks are defined. 24. Click [Save CS]. The default name is cs_c Click [Return] - [Return] - [Write] - [Write CS in f1]. The default name for the file is chemshifts_c13. 11