Nature Publishing Group

Transcription

1 Figure S I II III IV ratio ssdna (S/G) WT hr hr hr V VII X VI XI VIII IX ratio ssdna (S/G) rad hr hr hr 6 7 Chromosome Coordinate (kb) 6 6 Nature Publishing Group

2 Figure S Continued XII XIII ratio ssdna (S/G) WT hr hr hr XIV 6 7 XV ratio ssdna (S/G) rad hr hr hr XVI Chromosome Coordinate (kb) 6 Nature Publishing Group

3 Figure S. Overlay of smoothed ssdna profiles for S. cerevisiae. The time course for both WT and rad strains are shown in series of increasing color intensity. WT cell profiles at, and -hour post release are shown as light purple, magenta and dark purple curves, respectively; rad cell profiles at, and -hour post release are shown as yellow, orange and red curves, respectively. WT cell profiles are plotted on the Y (left) axes and rad cell profiles are plotted on the Y (right) axes. Positions of Pro- ARSs are shown as green diamonds. Positions of clustered origins (those that appear in at least two of the timed samples) from ssdna profiles of rad cells are shown as filled blue circles. Positions of singleton origins (those that appear in only one of the three timed sample) are shown as filled red circles. 6 Nature Publishing Group

4 Figure S I II III IV ratio ssdna (S/G) WT hr V VII VI VIII IX copy number X XI 6 7 Chromosome Coordinate (kb) 6 6 Nature Publishing Group

5 Figure S Continued XII XIII ratio ssdna (S/G) WT hr XIV 6 7 copy number XV XVI Chromosome Coordinate (kb) 6 Nature Publishing Group

6 Figure S. Comparison between S. cerevisiae Rad-unchecked origins and early origins identified by copy number change detection at 9 minutes in HU. The smoothed data of ratios of ssdna (S/G) for WT cells at hour post release (green curves) are overlaid with changes in copy number of genomic DNA in the presence of HU (orange shaded curves). The ratios of ssdna are plotted on the Y (left) axes and the copy number change is plotted on the Y (right) axes. The positions of all the clustered ssdna peaks in rad cells representing all origins are indicated by filled blue circles; the peaks that occur in a single timed sample are indicated by filled red circles. The ssdna peaks that meet the statistical criteria for local maxima in the WT profile are indicated by open diamonds: those that match the clustered ssdna peaks in rad cells are shown in blue and those that match the singleton ssdna peaks in rad cells are shown in red. The four ssdna peaks that only appear in WT cells but are not identified as clustered or singleton ssdna peaks in rad cells are shown as solid black diamonds. 6 Nature Publishing Group

7 Relative Ratio ssdna (S/G) Figure S_chromosome I 6 AT AT ORI pcr ORI 76 Chromosome I ORI c pars767 ORI c Chromosome Coordinate (kb) 6 Nature Publishing Group ORI 8 ORI AT AT rip ARS ORI 9

8 Figure S_chromosome II Chromosome II Relative Ratio ssdna (S/G) 6 AT AT 6 AT ORI c7 ORI 7 ARS 7 AT 8 tug/rum pars7 pars77 6 AT 96 pars77 6 Chromosome Coordinate (kb) 6 Nature Publishing Group

9 Figure S_chromosome III Chromosome III Relative Ratio ssdna (S/G) 6 ARS AT / AT ORI c nmt 6 8 Chromosome Coordinate (kb) pars7 6 Nature Publishing Group

10 Figure S. Overlay of ssdna profiles for S. pombe WT (purple) and cds (orange) cells. WT cell profiles are plotted on the Y (left) axes and cds cell profiles are plotted on the Y (right) axes. Positions of significant ssdna peaks in WT and cds cells are indicated by purple and orange filled circles respectively. Positions of AT-rich islands reported by Segurado et al. 9 are shown as green filled circles. Positions of previously mapped origins from Segurado et al. 9 and references therein are shown as red filled circles. Those previously mapped origins that have also been identified as significant ssdna peaks in cds cells are labeled above the graphs. The two origins that eluded our analysis are boxed. 6 Nature Publishing Group

11 SUPPLEMENTARY INFORMATION The data discussed in this publication have been deposited in NCBI s Gene Expression Omnibus (GEO, ) and are accessible through GEO Series accession number GSE99. Normalization Our experiments utilize Agilent microarrays that contain,87 oligonucleotide probes, covering 6,6 S. cerevisiae ORFs with extensive replicates and Eurogentec microarrays that contain, PCR amplified probes covering,976 S. pombe ORFs with replicates. In our microarray experiments, we sought a quantitative measurement of the ratio of single stranded DNA between a time point N (denoted TP N ) and a reference time point (denoted TP, for time point zero). After collection of DNA for TP N and TP and isolation of genomic DNA for time points N and, we differentially labeled DNA complementary to the single stranded DNA, using Cy- and Cy- dyes, respectively. The labeled samples were hybridized to the microarray. Using GenePix. software (Axon), we converted the Cy- and Cy- fluorescence intensities from TIFF files into numerical intensity data for each labeled sample and extracted the background subtracted median feature pixel intensity numbers for each spot on the array. Averaging over duplicated features, we obtained TP N Cy- signals (denoted a i *) and TP Cy- signals (denoted b i *), where i indexes the set of all yeast ORFs. We arrived at Equation (): observed ratio of ssdna at ORF i = a i */b i *, i indexing the ORFs. 6 Nature Publishing Group

12 This calculation led to the chromosomal observed single stranded ratio profile by plotting the data points (cj, a j */b j * ), where j runs over the ORFs for a given chromosome. Although informative, this profile suffers from the fact it is not quantitative: the TIFF files produced by core facilities arbitrarily adjust the gain in both Cy channels to achieve a roughly equal balance of total signal. To get around this problem, the data were normalized and made quantitative by applying the scheme laid out in Collingwood et al. (manuscript in preparation). The normalization uses an external measurement of ssdna content in S vs. G (slot blot) to correct one of the Cy channels, so as to restore to the microarray data the S/G ratio seen in the external calibration. The key result asserts the existence of a computable constant g, such that Equation (): actual ratio of ssdna at ORF i = g a i */b i *. The constant g=n/m, where n= (total TP N ss DNA)/(total TP ss DNA), m=(total Cy- signal)/(total Cy- signal). The constant m is directly computed from the array data. Because we isolated equal amounts of TP N and TP DNA, 6 Nature Publishing Group

13 n = (total TP N ssdna)/(total TP ssdna) = [(total TP N ssdna)/(total TP N DNA)]/ [(total TP ssdna)/(total TP N DNA)] = [(total TP N ssdna)/total TP N DNA]/ [(total TP ssdna)/total TP DNA] = (% TP N ssdna)/(% TP ssdna). We were able to experimentally compute this ratio of percentages, thus allowing us to compute the normalization constant g. We then obtained the actual single stranded ratio profile by plotting the data points (c j, g a j */b j * ), where j runs over the ORFs for a given chromosome. Smoothing We transformed the raw data of ssdna ratio by using the Fourier convolution smoothing technique previously introduced to obtain a smoothed profile. The smoothed profiles offered the advantage of prominently identifying local extrema in the data. In our application of smoothing, a window of kb was specified and a moving average with this window size was constructed. We used this moving average as a target and selected the closest Fourier smoothing among a large family of smoothings. See the supporting online text to Raghuraman et al. for full details of this procedure. Extrema detection Given any discrete dataset of points (x i,y i ), we detected local extrema as follows: First, for each data point, we calculated the numbers 6 Nature Publishing Group

14 S i L= (y i -y i- )/(x i -x i- ) and S i R= (y i+ y i )/(x i+ -x i ). If S i L > and S i R <, then we flagged the point (x i,y i ) as a local maximum, whereas if S i L < and S i R >, then we flagged the point (x i,y i ) as a local minimum. Identification of significant ssdna peaks In order to calculate standard deviation in the background level of ssdna labeling in all timed samples, we first identified and flagged those data points with values above the median value in each data set, thus removing prominent peaks from our estimation of background variation. We then removed those data points that were flagged in any timed sample data from all timed sample data in all further calculations. Averaging the three median values (Own Median), we obtained the Average Median and normalized the remaining data points by multiplying each value with the constant M (M=Average Median/Own Median). Standard deviation was calculated as the square root of variance (variance = Σ i {/[(X i, - X i, ) +( X i, - X i, ) +( X i, - X i, ) ] / total number of data points}, where i indexes the data points and the subscripts,, and index time points,, and hr). We present a sample calculation (numbers have been rounded here to two decimal places for clarity): 6 Nature Publishing Group

15 Calculate median for rad at, and hour Median_rad_hr =.9 Median_rad_hr =.8 Median_rad_hr =. Calculate average of the Medians above Average of Median =. Calculate the normalization factor (costant M) for rad at, and hr M_rad_hr =.9/. =.8 M_rad_hr =.8/. =. M_rad_hr =./. =. Normalize each data set by multiplying the smoothed ratio of ssdna (e.g., rad_xhr in the spread sheet below) with the normalization factor constant M and arrive at a normalized value (e.g., rad_xhr_nm in the spread sheet below). Shown below is a spreadsheet of calculations for a portion of chromosome. chr coord rad_hr rad_hr_nm rad_hr rad_hr_nm rad_hr rad_hr_nm Nature Publishing Group

16 Calculate variance and standard deviation Variance = Σ i {/[(X i, - X i, ) +( X i, - X i, ) +( X i, - X i, ) ]} / total number of data points, where i indexes the data points and the subscripts,, and index time points,, and hr). Standard deviation = square root of variance For example, at the first coordinate (8 kb on chr ), calculate /[( ) +(.6-.6) +( ) ] =.7. Repeat this calculation for all the coordinates in the genome and average all of them and arrive at Variance =.8; Standard deviation =.6. We next identified the local maxima and minima in each timed sample and calculated the difference between every local maximum and its two flanking minima (note that for some of the telomeric points, not every local maximum is flanked by two 6 Nature Publishing Group

17 local minima). Those local maxima with values that are above standard deviations from both its flanking local minima were considered significant ssdna peaks. Those telomeric local maxima that only have a single flanking local minimum but are above standard deviation of the said local minimum are also considered significant ssdna peaks. 6 Nature Publishing Group

18 Table S. Complete list of clusters of origins identified from the ssdna profiles of rad cells in HU. Column contains the chromosome number. Columns to list locations of all elements of each clustered origin from ssdna profiles of rad cells at three time points as described in the text. For those origin clusters that contain fewer than three elements, blank entries indicate that no significant peak was identified at the given location in that sample. Peak locations that were identified only once among all repetitions of the experiment (true singletons) are shown with asterisks. The positions of ssdna peaks that appear in the WT hour sample are listed in column. The 7 ssdna peaks in WT that do not match clustered ssdna peaks in rad cells are italicized. Note that of these 7 peaks in WT do match a singleton ssdna peak in rad cells. The name and position of the corresponding Pro-ARS are adapted from Wyrick et al., and are listed in column 6 through 9. False indicates predicted Pro-ARSs found not to show ARS activity. Chromosome Position Position Position Position Pro-ARS/ Start (kb) End (kb) Mid(kb) rad_hr rad_hr rad_hr WT_hr ARS (kb) (kb) (kb) (kb) name * 6 Nature Publishing Group

19 * * * / FALSE FALSE Nature Publishing Group

20 6 6 7 FALSE * Nature Publishing Group

21 * * / FALSE Nature Publishing Group

22 6 FALSE Nature Publishing Group

23 * FALSE Nature Publishing Group

24 FALSE FALSE FALSE FALSE FALSE Nature Publishing Group

25 * Nature Publishing Group

26 * * Nature Publishing Group

27 * * Nature Publishing Group

28 * * Nature Publishing Group

29 Table S. List of ABOs that do not match the clustered ssdna peaks from the rad ssdna profiles within a kb distance. The chromosomal coordinate positions for Pro- ARSs and origins from previous studies are imported from MacAlpine and Bell, available at Column is chromosome number. Column is the calculated average position of the following positions reported in MacAlpine and Bell : the start of Pro-ARSs from Wyrick et al. (column ), the origin positions from Yabuki et al. (column ) and the origin positions from Raghuraman et al. (column ). The comparison between ABOs and our clustered ssdna peaks uses column as the average position of an ABO. chr Ave_coord(kb) Pro-ARS_start(kb) CN_start(kb) HL_start(kb) Nature Publishing Group