gada R package: User`s manual
Contents
1. 0 SSS 18 e Se eS 5 MI 274 Genomic Position 247189096 Figure 5 Gains red colors and loses blue colors relative frequencies for 8 individuals from general popu lation along the entire genomeo gains and loses corresponding to whole genome analyisis while Figure 6 shows the same information for chromosome 6 They can be obtained by typing gt plotWG allSamples and gt plot allSamples 6 show ind TRUE The parameter specifies whether individuals are separated When a large number of individuals is analyzed it is recommended not to change the default parameter 18 Chromosome 6 8 samples 1 i i 1 1 1 i 1 1 1 1 1 1 1 i 1 1 1 i i i 1 1 1 1 1 E i T T 1 i 1 i 1 1 1 1 1 i 1 1 1 i i 1 1 5 8 5 ovog Idures IAIpu sejdures2. 0 0 Report our 0 0092 0 3314 0 0006 0 3200 04 NaN Nc NaN NC Nat O Locus Summary 0 0000 0 9903 0 0023 0 8430 0 0 B 0 9805 0 0537 0 9654 0 0209 09 1 13 Locus x DNA 0 0000 0 3494 0 0070 0 6267 0 0 020 0 4081 0 2095 0 0033 0 1156 NC 04 B 0 8748 0 2165 0 9559 0 3537 09 7 757 740 055 0 3800 0 2869 NC 0 3813 0 3286 0 9 Norm Theta 0 5794 0 3266 0 9951 0 1420 BB 09 B 0 9549 0 1783 BB 0 9622 0 0904 0 9 E 0 3704 0 5582 NC 0 3806 0 3888 NC 03 Samples Table 0 0069 0 3959 0 0065 0 1530 0 0 0 4150 0 1414 AB 0 4308 0 3163 BB 09 B 0 9759 0 2405 0 9648 0 4038 BB 09 B 0 9870 0 1161 0 0000 0 1652 0 0 NaN NC NaN Nat B 0 9919 0 0012 0 9809 0 0825 BB 09 B 0 9253 0 0011 0 9339 0 1344 BB 09 NC NaN NaN NC Nat 4424530408 RO2CO1 0 9068455 Unknown 57 300 0 3315 0 4316 NC 0 3537 0 2065 4424530408 RO2CO2 0 9098104 Unknown 46 301 M 4424530409 RO1CD1 0 904660 Unknown 65 22 gt 4424530409 RDICO2 0 9030093 Unknown 58 228 4424530409 802601 0 8970144 Unknown 88 289 4424530409 RO2CO2 0 9026254 Unknown 87 298 1 gt x 4424530009 ROICO1 0 9066092_ Unknown 75 260 Cancel Back Next gt Finish 4424530009 ROICO2 0 9048520 Unknown 68 244 4424530009 _202 01 0 907147 Unknown 1
3. 0 9160 0 0141 0 8644 0 9160 0 0172 1 0040 AA Name MUN lt gt 1296 19 22101812 NC 0 0000 0 5942 0 4450 NC 0 0000 0 5772 0 4433 NC Samples Table 1147 19 593217 0 4183 0 1394 1 3735 0 2758 0 5038 1 4822 Chr 1296 19 6185618 0 6181 0 0839 1 0172 0 7224 0 0320 1 1224 m E an Position 1296 6 106167295 BB 0 9047 0 9646 1 2866 0 9047 0 5421 1 5029 4424530002 RO1CO1 3461 6 130055946 0 7919 0 0230 1 0513 0 7919 0 0343 1 0343 4424530002 RO1CO 6 1456 19 48370536 BB 0 2860 0 8757 0 4276 BB 0 8508 0 9223 0 3565 BB 4424530002 RO2COI 7 1296 19 59850976 0 8838 0 0197 0 8201 0 8838 0 0319 0 8086 Sample ID 4424530002 R0202 B 2162 19 39735732 0 0000 0 0134 1 0245 NC 0 0000 0 0160 1 0266 4424530408 RO2COL 4424530068 01 co 9 1296 19 18265148 0 8624 0 9813 1 2072 0 8624 0 9861 1 1534 BB 4424530408 _ 0202 enum Hi ubeolumns 31 4424530409 01 01 Displayed Subcolumns Ven 8 4424530409 RO1CO2 GType 00 573397 Fiter Fiteris not active 4424530409 RO2CDI Y 4424530409 RO2CO2 Top Alleles dbx 4424530009_R01CO1 4424530009 RO1COZ 1 E Y oe 4424530009 RO2COI Orig all 4424530009 80202 45 4424530410 01 01 ee pana ae ibe ee 4424530410 RO1CO2 Sc
4. 0 02312321 Array 1 done Array 2 The estimated sigma2 0 01455486 Array 2 done Array 3 The estimated sigma2 0 01264846 14 Array 3 done Array 4 The estimated sigma2 0 01334702 Array 4 done Array 5 The estimated sigma2 0 01252028 Array 5 done Array 6 The estimated sigma2 0 02356903 Array 6 done Array 7 The estimated sigma2 0 02532079 Array 7 done Array 8 The estimated sigma2 0 02927364 Array 8 done Segmentation procedure for 8 samples done In this case we perform the segmentation procedure for all samples in the folder SBL that have been It is possible to perform segmentation procedure for a subset of imported as a setupGADA objects individuals by using the argument Samples as following gt Not run gt parSBL myExample Samples c 4 8 estim sigma2 TRUE gt End not run The SBL result for each array is stored in a directory called SBL Notice that in this case the argument Samples is a vector pair indicating the first and the last individual to be analyzed Therefore if the process is stopped for any reason the analysis for the subjects left out is easily resumed Similarly the backward elimination BE for multiple individuals is implemented in the function multiBE gt parBE myExample T 8 MinSegLen 8 Retrieving annotation data Backward elimination procedure Array 1 Sparse Bayesian Learning SBL Backward Elimination proced
5. 103 rs964927 4 6 79070425 losses probe Freq chr pos 1 cnv30178p1 2 6 30311634 2 cnv30178p3 2 6 30312496 3 cnv30180p1 2 6 30320931 4 cnv30180p2 2 6 30321486 5 cnv30180p4 2 6 30322754 21 cnv30813p1 2 6 32066939 20 22 cnv30813p3 2 6 32067243 23 cnv30813p5 2 6 32067459 24 cnv30814p12 2 6 32068749 25 cnv30814p18 2 6 32069461 26 cnv30814p4 2 6 32067857 27 30815 1 2 6 32069619 28 30815 12 2 6 32070703 29 30815 18 2 6 32071423 30 30817 1 2 6 32073734 Notice that this function requires the argument chr corresponding to a desired chromosome By default this function only returns those probes that are altered gains and losses in two different data frames in more than 10 of samples This can be changed by using the argument min perc For example probes2 lt getAlteredProbes allSamples chr 6 min perc 0 50 will return the probes that present a gain or a loss in more that 50 of individuals Another useful tool is the function exportToBED to export data gt exportToBED allSamples File BED txt has been generated at home jrgonzalez CREAL GADA This function generates a file called BED txt that contains the required information to be displayed in most popular genome browsers UCSC http genome ucsc edu ENSEMBL http www ensemb1l org index html 4 Multivariate analysis of segmented data As an illustration of the type of analysis that can follow data segmentation we show the multivariate
6. discrimination of three HapMap populations 90 CEU 90 YRI 45 45 JPT CEL files for each of the samples are available at http www hapmap org We used Aroma Affimetrix for its normalization and GADA for its segmentation as described in the previous sections We have implemented a set of functions to reduce the data set without much loss of information and to perform a multi class discrimination and variable ranking The output of GADA for a set of subjects is an object of class ParGADA that can be used to build a matrix which encodes the copy number status of each probe and subject in the sample Performing a segmentation for each separate populations produces a ParGADA object for each group which can be recovered and summarized accordingly change to CEU directory ParAffyData lt setupParGADAaffy log2ratioCol 4 NumCols 4 Samples CUE lt summary ParAffyData length c 500 6e9 change to YRI directory ParAffyData lt setupParGADAaffy log2ratioCol 4 NumCols 4 Samples YRI lt summary ParAffyData length c 500 6e9 change to CBH JPT directory ParAffyData lt setupParGADAaffy log2ratioCol 4 NumCols 4 Samples CJ summary ParAffyData length c 500 6e9 The matrix of segment callings for all probes and subjects is obtained with the function getReducedData Its arguments are the concatenation of the segmentation results in a single list segments lt c Samples CUE Samples YRI Samples CJ
7. sjen Figure 6 Gains red colors and loses blue colors for 8 individuals from general population on chromosome 19 The user can also obtain those probes that are altered gains losses in a given proportion of individuals This can be done using the function getAlteredProbes as following gt probes lt getAlteredProbes allSamples chr 6 gt probes gt probes gains probe Freq chr pos 3 rs1064611 2 6 32630503 4 rs1093580 4 6 79056617 6 rsi1757159 2 6 32628250 8 rsii1759557 2 6 32628011 9 rsi11964123 4 6 79052979 14 rs16889854 4 6 79081009 15 rs16889859 4 6 79082584 31 rs28490179 2 6 32626983 37 rs28880026 2 6 32625376 47 rs34182525 2 6 32631416 48 534781832 2 6 32628606 49 rs34867789 2 6 32629229 rs3819713 2 6 32624385 62 rs6911209 4 6 79065940 65 rs6918807 4 6 79063712 67 rs6931912 4 6 79065999 68 rs6932920 4 6 79059458 69 rs7749022 4 6 79075016 70 rs7TT3124 4 6 79090197 71 rsTTT4454 4 6 79077999 72 rs818251 2 6 79031111 73 rs818253 3 6 79031809 74 rs818258 4 6 79034386 75 rs818262 4 6 79036117 76 rs818280 4 6 79088461 77 rs818284 4 6 79083083 78 rs818285 4 6 79083049 79 rs818288 4 6 79078423 80 rs818290 4 6 79077158 81 rs818295 4 6 79069278 82 rs818301 4 6 79056822 83 rs818310 4 6 79042356 84 rs818313 4 6 79039487 94 rs9361392 4 6 79067895 97 rs9443550 4 6 79083326 98 rs9448350 4 6 79069674 99 rs9448356 4 6 79076024 100 rs9448357 4 6 79076473 101 rs9448361 4 6 79086086
8. t ae ale eee BO reu 12 3 1 1 Importing a collection of Illumina array data 12 3 1 2 Importing a collection of Affymetrix array data 14 3 2 Segmentation procedure 14 3 21 Paxalell segmient tion 2 OX tus 16 3 3 S mmaryzing results X sears erbe lee RE ATHEN Quis ea eae Shap et 16 4 Multivariate analysis of segmented data 21 Association analysis 24 6 Exporting data from Illumina and Affymetrix platforms to gada 26 6 1 Exporting data from Bead 26 6 2 Exporting data from Affymetrix genotyping console GTC 31 6 3 Exporting data from Affymetrix power tools APT 32 7 Tutorial session with Affymetrix data 34 7 1 Analyzing a single Affymetrix array 34 7 2 Analyzing a collection of 90 Affymetrix 38 8 Connection with Aroma Affymetrix 45 1 Installation The scripts and package described in this manual are accessible at http groups google com group gadaproject where the tar gz source file can also be found From the R command line gada is installed using gt install packages gada 0 8 0 tar gz repos NULL Another option is to install the package from CRAN not yet available gt install packages gada Then the package is loaded by typing gt
9. 17 1 422314e 14 4 BlkCnv32Chri2 7 061229e 16 2 600298e 13 BlkCnv31Chri2 7 421259e 15 2 186303e 12 6 BlkCnv41Chri7 1 455019e 14 3 572072e 12 7 BlkCnv30Chri2 9 261514e 14 1 948887e 11 8 BlkCnv40Chri7 1 039148e 12 1 913331e 10 9 BlkCnv33Chri2 1 187513e 12 1 943564e 10 10 BlkCnv39Chri7 6 314224e 11 9 300852e 09 11 BlkCnv80Chr8 1 052788e 10 1 409779e 08 12 BlkCnv34Chr4 1 230331 10 1 510231e 08 13 BlkCnv81Chr8 6 391850 10 7 242458e 08 14 BlkCnv78Chr8 8 311588e 10 8 744977e 08 15 BlkCnv83Chr8 9 031225e 10 8 868663e 08 16 BlkOnv4Chr2 1 542300e 09 1 135904e 07 17 BlkOnv7Chr2 1 342954e 09 1 135904e 07 18 BlkCnv21Chr3 1 523509e 09 1 135904 07 19 BlkCnv22Chr3 1 523509e 09 1 135904e 07 20 BlkCnv79Chr8 1 407583e 09 1 135904e 07 Finally a plot with 10510 p values can be obtained see Figure 8 by executing plot ansO cnv blocks cex 1 plot ans cnv blocks cex 1 24 logio P Crude analysis S wo m m m 1 1 P d a n 2 05 Li 4 Li logio P T T lj Fk ok 34 5 6 7 12 14 Chromosome 17 20 15 10 Adjusted for population two first eigen values form non supervised MCA T T T T TT IT 3 4 5 6 7 9 12 14 17 Chromosome Figure 8 Association analysis for 1 473 blocks for HapMap data using simulated case control status 25 6 Exporting data from Illumi
10. 508 510 512 514 laboratories etc IniProbe 742429 25179149 25264951 41725184 41727031 50489793 52878645 70694440 80309805 80329403 82193305 97357234 110070351 115538986 125716562 126565408 128403153 130307929 137853309 144181999 147115894 154545424 25178194 25264714 41720331 41726332 50481809 52876849 61404617 80309386 80320023 82192926 97356749 107627389 110984890 115553916 125719857 126586900 128461071 130873100 138032698 144968405 147246249 154871186 4906 41 2578 3 1289 200 1970 1612 4 431 2802 1695 244 Mo 60 20 111 10 7 The function estimates the reference MeanAmp chromosome State 033250509 151262862 037778763 355878820 023630115 029492261 020525958 036965666 327258580 045561250 012820850 043190020 054825800 133537435 591493700 768452000 012350443 346408771 052194362 292001868 411471160 018561945 1 HG HH H RP RB P4 24 ratio corresponding to two copy numbers Base Amplitude of copy number 2 in the output computing the median intensity along the autosomal genome This value can also be manually specified using summary step2 BaseAmp 0 After that the segment mean amplitude MeanAmp is normalized by subtracting the reference ratio of two copy numbers in order to take into account differences between arrays with respect to uncontrolled
11. Figure 20 Gains red and losses blue frequencies for 90 CEU individuals along the entire genome 42 10 CN_146310 CN_146322 CN_146343 243 SNP_A 8534896 246 SNP_A 8658169 losses probe 8 CN_146343 12 CN_429094 15 CN_739260 16 CN_739262 17 CN_739264 91 CN_751784 92 CN_751786 185 SNP_A 4288097 199 SNP_A 8505450 62 17 62 17 27 17 62 17 35 17 Freq chr 16 17 18 17 18 17 21 17 21 17 17 17 15 17 17 17 21 17 41623467 41667651 41750175 41581663 41522088 41750175 41750177 41750183 41756820 41764411 18387392 18394150 18308103 41927619 Finally the function exportToBED can be used to save the CNA segments in BED format gt exportToBED allSamples File BED txt has been generated at data cluster1 rpique datasets aptDataNew AffyCeuGW6GTC301 which are stored on the BED txt file head BED txt chri chri chri chri chri chri chri chri chri chri 72541512 147303136 147442911 147509275 147526028 246815805 72541512 105820716 110027431 111179076 72583724 147438362 147496455 147521544 147703454 246877269 72583709 105823886 110044464 111189737 06985 06985 06985 06985 06985 06985 NA06991 NA06991 NA06991 NA06991 300 300 300 300 300 300 300 300 300 300 72541512 147303136 147442911 147509275 147526028 246815805 72541512 105820716 110027431 111179076 72583724 147438362 147496455 147521544
12. Name Chr Position GType Allele Freq Log R Ratio rs1000050 1 161003087 AB 0 4960448 0 1494603 151000073 1 155522020 AB 0 4824853 0 00509767 151000313 1 15278076 0 0 1521843 151000476 1 58694104 AA 0 001480275 0 09277323 151000533 1 166549115 0 5048196 0 002900129 151000543 1 242254223 BB 1 0 01190711 rs1000730 1 230030224 AB 0 5039815 0 06360321 rs1000731 1 230030114 BB 0 998738 0 05265531 rs1000997 1 15998548 AB 0 5515608 0 03962962 rs1001149 1 150775186 BB 1 0 06456274 rs1001160 1 76131179 AA O 0 0726123 rs1001193 1 145633001 AA O 0 1115284 and can be downloaded with gt download file http www creal cat jrgonzalez GADA datalllumina txt dataIllumina txt trying URL http www creal cat jrgonzalez GADA datalllumina txt Content type text plain length 24698671 bytes 23 6 Mb opened URL downloaded 23 6 Mb The second example is sample obtained from the Affymetrix plataform head 500 NA06985_GW6_C MyTest CN5 CNCHP txt Comments Comments Comments ProbeSetName Chromosome Position CNState Log2Ratio SmoothSignal LOH Allele Difference CN_473963 1 51586 2 0 257667 1 054558 nan nan CN 473964 1 51659 2 0 264712 1 054389 nan nan CN 473965 1 51674 2 0 043675 1 054354 nan nan CN 473981 1 52771 2 0 402939 1 051817 nan nan CN 473982 1 52788 2 0 134605 1 051777 CN_497981 1 62627 2 0 006367 1 029375 nan nan CN 502615 1 75787 2 0 677508 1 000571 nan nan CN 502613 1 75849 2 0 3
13. factors like amount of DNA different The segments are then classified as Gain State 1 Loss State 1 or Neutral State 0 depending on whether the segment mean amplitude is above below or non significantly different than BaseAmp Only the segment with significant deviations gains or losses are reported by summary plotRatio shows the log ratio intensities as well as the segments obtained after backward elimination procedure Figure 2 3 gt plotRatio step2 Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 4 5 and minimun length size 3 Number of segments Base Amplitude of copy number 2 chr 1 22 0 0174 X 0 221 Y 0 0275 This plot can also be obtained chromosome wise As an example Figure 2 3 shows the intensities and segments found after applying the backward elimination procedure in chromosome 12 log ratio 9 10 11 12 18 14 15 16 17 18 19 20 2122 X Y Chromosome Figure 3 log ratio intensities and segments for the entire genome gt plotRatio step2 chr 12 Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 4 5 and minimun length size 3 Number of segments 515 Base Amplitude of copy number 2 chr 1 22 0 0174 X 0 221 Y 0 0275 10 Chromosome 12 02 045 06 Figure 4 log ratio intensities by chromosome and break points for chromosome 12 11 3 Multiple array analysis T
14. 0 3537 BB 0 9 Standard Format Options 4 0 3852 0 2597 0 3800 0 2869 NC 0 3813 0 3286 0 9 80 0 5532 0 0018 0 5794 0 3266 BB 0 9951 0 1420 BB 0 9 Displssed retis 940 0 9531 01399 BB 0 9549 01783 BB 0 9622 0 0904 09 SNP Name al 40 NC 0 3583 0 6271 NC 0 3704 0 5582 NC 0 3806 0 3888 NC 0 3 Sample 1D R 107 0 0088 0 2326 0 0069 0 3959 0 0065 0 1530 0 0 750 0 4683 0 2966 0 4150 0 1414 0 4308 0 3163 BB 0 9 Y 31 BB 1 0000 0 3290 BB 0 9759 0 2405 0 9648 0 4038 0 9 3 XRaw 79 0 9833 0 1023 0 9870 0 1161 0 0000 0 1652 0 0 S Raw 4000 NC NaN NaN NC NaN NC Nat B Allele Freq 090345 0 9839 0 0501 0 9919 0 0012 0 9809 0 0825 BB 0 9 NM 49 BB 0 9286 0 1625 BB 0 9253 0 0011 0 9339 0 1344 0 9 33000 RUTCUZ FAT i 2379 NaN NaN NC NaN NC Nat pod 54465 NC 0 3478 0 3347 NC 0 3315 0 4316 0 3537 0 2065 9 3 4424530409 01 01 Group by sampe O SNP 3 4424530409 RO1COZ 897 Sell Fiter Fiter is not active 4424530409 RO2COI CERT avorite Format 4424530409 RD2CO2 Tuus 0px 4424530009 01 01 Om Default 4424530009 01 02 T 4424530009 02 01 Estimat 256 4 MB 4424530009_ROZCO2 BRSCUIS A Sze Parenti Parenti e child ChidiRe Parenti R
15. 0 8S2Cpaired GenomeWideSNP 6 gada N788 chr22 x0016 png Mozilla Firefox File Edit View History Bookmarks Tools Help 0 ie maatarcustertnpique datasetsireports MedulowithContro s ACC 2CraN2C XY R2CAVGH2C 2BS0ON2CANZBBN2CFLNKZE 2 v gt 6 7 openSUSE Getting Started Gj Latest Headlines Chromosome Explorer 2006 2008 Bengusony Help About 2 bubo 32 4 128 Sample 788 5 Sets gada Samples NZBG N799 N790 N791 N792 N793 N794 N795 J N796 N797IN798 N799 104802 04803 0480404805 04806 04807 11 048124813 Step prev next Play start stop slower fasten Chromosome 017 0 23102 10 10 0 077 087 09 10 11 12013702 4 15 16 177 1870197201 2209 iol S Physical position Identified regions GenomeWideSNP Gioadaegiong xis Image URL Done Figure 22 Browsing the segmentation results aroma affymetrix Chromosome Explorer Alternatively the segments can be manually extracted to use in downstream analysis using gt cnrs lt getRegions gada arrays 1 chromosomes 1 verbose verbose 46 Extracting regions from all fits Obtaining CN model fits or fit if missing Obtaining CN model fits or fit if missing done Extracting regions for chromosome 1 Extracting regions for chromosome 1 done Extracted regions data frame 22 obs of 5 variables chromosome int 1111111111 start num 51599 62
16. 00 33457952 4424530009 80202 0 9045697 Unknown 90 2894 6488 28 Error chidjRe Paren an Parent2 4424530410 ROCO1 0 8942768 Unknown 81 2045 538 19 Index p Lis Child Rep me Rep Parent Rep index P 4424530410 RO1CO2 0 8973569 Unknown 6 2087 5482 18 Index GType 4424530410 ROZCOL 0 8923424 Unknown 112 2442 6287 2 l 4424530410 RD2CO2 0 8964060 Unknown 91 2636 65 24 lt gt Rows 0 0 Sel 0 FilerFilter is not active Figure 11 Exporting log2ratios from BeadStudio tool in different files 29 BeadStudio Genotyping WGG0014 gdt Analysis Tools Window DSO E SNP Graph gt Full Data Table SNP Table Paired Sample Table 1b x amp 1510419363 2 Allele Log BAlele Log g BAI L Name Chr Postion Ratio 6796 Freq Ratio Freq mato 6796 rn zo o 710419302 19 3887448 AB 0 4613 0 0526 0 4856 0 0140 AB 0 4587 0 0083 151041944 1 234778604 AA 0 0040 0 3532 0 0023 0 1177 0 0080 0 1251 AA 0 0 160 F 1510419472 19 13405806 0 9791 0 4932 BB 0 9827 0 4606 1 0000 0 4915 10
17. 00 items Array 7 done Importing array sample C8 Read 5364100 items Array 8 done Applying setupGADAIllumina for 8 samples done Creating objects of class setupGADA for all input files done This function calls repeatedly the function setupGADAIllumina so the arguments log2ratioCol and NumCols are passed through the function setupGADAIllumina see section 2 1 Other arguments for setupGADAIllumina can also be set from this function The function saves an object of class setupGADA for each sample in the directory SBL The function returns an object of class parGADA which allows the process to be resumed later object of class parGADA contains this information gt myExample 1 home jrgonzalez CREAL GADA 13 attr class 1 parGADA attr type 1 Illumina attr labels samples 1 Ci C2 C3 c4 C5 C6 c8 attr Samples 1 8 This object is throughly used in the analysis and plotting procedures For instance a plot for individual 4 with log2ratio intensities can be obtained with gt plot for sample 4 gt plotRatio myExample Sample 4 and the same plot including the segments is obtained via gt plot for sample 4 with segments gt plotRatio myExample Sample 4 segments TRUE It is recommended to save this object to continue performing the analysis in case of R need to be restarted gt save myExample file myExample Rdata 3 1 2 Importing a colle
18. 0022 13 99259220 AA 0 2494328 0 rs10000272 4 189927377 0 1513728 0 002721502 26 Beadstudio Genotyping 1660014 Ele Edi Analysis Tools Window SNP Graph 4 gt X Full Data Table stie Table Paired Sample Table 4 332 Ei B 4 1510419302 Index Name Address Chr Position GType Score Theta R GType Score Theta 32917 1510418175 1447 19 51382464 NC 0 0000 0 5966 2 1366 0 0000 0 5944 2 1295 NC 32918 510418205 2160 19 6051433 BB 0 8067 0 9398 1 3846 0 8067 0 5136 1 5346 32919 1510418296 2260 19 58594425 0 6280 0 1068 0 5859 0 6280 0 1239 0 5233 BB 32920 151041832 2360 21 37104999 BB 0 9610 0 9895 1 0907 AA 0 9610 0 0131 1 2942 32921 1510418352 2460 19 17915851 0 9230 0 0430 0 7142 0 9230 0 0164 0 7291 32922 151041851 4267 7 142101110 NC 0 0000 0 9549 1 6342 NC 0 0000 0 9642 1 5882 NC 32923 51041856 2560 14 21080800 0 8840 0 0070 0 8969 0 8840 0 0198 0 9075 c 32924 1510418577 1296 19 34217651 AA 0 9220 0 0138 1 1011 AA 0 9220 0 0272 1 1107 E 32925 15104186 2660 5 145192615 AA 0 8237 0 0219 1 6605 AB 0 8237 0 3813 2 0576 5 32
19. 10000 iterations and change 7 17238078706828e 07 3 In FUN 1 24 24L 38 SBL algorithm did not converge after 10000 iterations and change 1 11067555152999e 08 gt After the SBL we continue with the BE step using parBE gt parBE ParAffyData T 6 MinSegLen 8 Retrieving annotation data done Backward elimination procedure for 90 samples Array 1 ora Sass ses assa srr ssa Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 6 and minimun length size 8 Number of segments 516 Base Amplitude of copy number 2 chr 1 22 0 0027 X 0 0343 2 2033 Array a SS eee Se Soe EI Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 6 and minimun length size 8 Number of segments 419 Base Amplitude of copy number 2 chr 1 22 0 0031 X 0 014 Y 2 1485 Array 90 soa ES Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 6 and minimun length size 8 Number of segments 451 Base Amplitude of copy number 2 chr 1 22 2e 04 X 0 0212 1 9905 Backward elimination procedure for 90 samples done The result of the segmentation of all the samples can be summarized by gt allSamples lt summary ParAffyData length c 500 6e9 gt gt print allSamples NOTE 2561 segments with length not in the range 500 6e 09 bases and with mean log2ratio in the range 0 28 0 16 have been discarded Number
20. 11041951 18 53391503 0 4501 0 1131 0 4391 0 0321 0 9618 0 1643 oa 110419562 19 16084287 0 0225 0 2398 0 0221 0 1776 0 0290 0 4491 0 0 rs0419669 19 49989909 0 9632 0 1381 BB 0 9556 0 1512 0 9702 0 1772 0 9 c 1207 e 1510419687 19 17650409 AB 0 5319 0 0818 AB 0 5528 0 1059 AB 0 5339 0 0780 00 iE 151041973 2 102321900 0 0000 0 7353 NC 0 4060 0 4918 0 8714 0 0665 NC 0 4 5 1510419759 19 52547139 0 0060 0 1066 0 0030 0 1881 0 0000 0 1136 0 0 z 1510419771 19 50275229 0 9794 1 1084 0 9969 1 1695 0 9741 1 1553 BB 09 8763 0 0105 0 7187 0 0017 0 6188 0 5974 0 1900 0 5 13636 BB 0 9970 0 6725 BB 0 9875 0 8038 1 0000 0 7338 0 4 1211 AA 0 0000 0 2492 AA 0 0060 0 2526 AA 0 0057 0 0934 AA 0 0 Genotyping Report 60 0 0114 0 3265 0 0092 0 3314 AA 0 0006 0 3200 AB 04 How would you like to format your Final report 360 NC NaN NC NaN NC NaN NC Nat 95 0 0002 0 9764 0 0000 0 9903 0 0023 0 8430 0 0 074 BB 0 9604 0 0430 BB 0 9805 0 0537 0 9654 0 0209 0 9 0702 0 0048 0 4399 0 0000 0 3494 0 0070 0 6267 0 0 Standard O Matrix O 3rd Party 47489 Nc 0 4087 0 0665 0 4081 02095 0 0033 0 156 NC 04 se BB 0 9501 0 3458 0 8748 0 2165 BB 0 9559
21. 147703454 246877269 72583709 105823886 110044464 111189737 255 0 0 255 0 0 255 0 0 255 0 0 255 0 0 255 0 0 0 255 255 255 0 0 0 255 255 255 0 0 then we can visualize this file on the UCSC genome browser http genome ucsc edu Figure 21 43 chri7 41500000 42000000 42500000 User Supplied Track NA12005 NA12874 07348 12892 NA12044 BRSSESSSSSESS3 S333 333333253322 NA10855 NA10839 NA10835 NA07019 NA12891 NA12878 NA12873 NA12865 12812 12762 12751 12740 12717 12707 NA12264 NA12248 NA12239 NA12057 NA12043 NA11994 NA11993 NA11881 NA11839 NA11832 NA10863 NA10860 NA07345 07055 07048 07022 NA10854 NA12144 NA10847 NA12753 NA12864 11995 NA06994 NA12872 NA12813 NA12761 NA12752 NA12236 NA12234 NA11992 11882 10859 10846 10830 10831 12006 06993 12155 12815 NA12763 NA12156 06985 12146 12707 NA07055 NA12248 NA12891 NA12878 12875 12874 12865 12751 12717 12239 NA11993 NA11829 NA10860 NA10838 NA07348 NA06991 NA12812 NA12154 NA12760 NA12057 07000 11831 sed on RefSeq UniProt GenBank CCDS and Comparative Genomics 513 RLI7ZH WNT3 ed CDC27 CRHR1H Al CRF 12 0855 2 1 CF NPEPPS H NPEPPS H ARL17P1 J 4 ARL17P1 LRRC37A LF Hi Lek 4 1 AA 12
22. 3 Chromosome 18 9 0 9 Chromosome 19 18 2 16 Chromosome 20 11 0 11 Chromosome 21 6 0 6 Chromosome 22 16 3 13 If for instance we are only interested in altered segments of size between 500 and 10 pair of bases we execute gt allSamples lt summary myExample length base c 500 10e6 Notice that this function only reports those segments with a mean log2ratio outside the given limits In this case these limits are 0 16 0 18 that is assumed to be the interval for segments with 2 copies By default these limits are estimated using a threshold approach to classify segments into Gain and Loss state The threshold is automatically estimated using the X chromosome of a normal population that includes males XY and females XX These limits can be changed by the user by changing the argument theshold As an example gt limits lt c 0 3 0 2 gt allSamples 2 lt summary myExample length base c 500 10e6 threshold limits There is a set of functions to simultaneously visualize gains and loses for each individual at both genomic and chromosome level They are accessed by the generic plot function We can plot information for all individuals in a the same figure using the generic function plot and the function plotWG Figure 5 shows 17 E Gains Losses CNV frequency summary 8 samples 17S ee eS 2 fo
23. 43111 1 000438 nan nan CN 502614 1 76175 0 1 440673 0 999744 nan nan CN 502616 1 76192 0 2 477916 0 999708 CN_502843 1 88453 2 0 135097 0 974336 nan nan CN_466171 1 218557 2 0 030157 1 597003 nan nan CN 468414 1 218926 2 0 475484 1 597018 nan nan CN 468412 1 219009 2 0 045742 1 597021 nan nan 1 219024 2 0 050614 1 597022 nan nan CN 468413 gt download file http www creal cat jrgonzalez GADA NA12248 GW6 C MyTest CN5 CNCHP txt NA12248 GW6 C MyTest CN5 CNCHP txt Both setupGADAIllumina or setupGADAAffy functions have the same arguments and are called in a similar manner gt dataIllumina lt setupGADAI1lumina file dataIllumina txt log2ratioCol 5 NumCols 6 Read 3367818 items and gt dataAffy lt setupGADAaffy file NA12248_GW6_C MyTest CN5 CNCHP txt NumCo1s 8 10og2ratioCol 5 Read 14507536 items where file indicates either the url or the path to the file which contains the data file dataIllumina txt log2ratioCol informs which column contains the log ratio instensities and NumCols gives the number of columns the file has Note that results exported from APT tools Section 6 3 should be specified with NumCols 8 log2ratioCol 5 Whereas those of Affymetrix Genotyping Console Section 6 2 are NumCols 4 and log2ratioCol 4 The argument sort is equal to TRUE by default in order to ensure that the data is correctly arranged by chromosomal position However if data is already order
24. 468412 1 219009 2 0 045742 1 597021 nan nan CN 468413 1 219024 2 0 050614 1 597022 nan nan 33 7 Tutorial session with Affymetrix data 7 1 Analyzing a single Affymetrix array The data used in this example can be downloaded from gt download file http www creal cat jrgonzalez GADA NA12248_GW6_C MyTest CN5 CNCHP txt NA12248_GW6_C MyTest CN5 CNCHP txt trying URL http www creal cat jrgonzalez GADA NA12248 GW6 C MyTest CN5 CNCHP txt Content type text plain length 97542566 bytes 93 0 Mb opened URL downloaded 93 0 Mb A single Affymetrix array can be imported to gada by executing gt dataAffy lt setupGADAaffy NA12248 GW6 C MyTest CN5 CNCHP txt NumCols 8 log2ratioCol 5 Read 14507536 items We use NumCols 8 log2ratioCol 5 for the results exported from APT tools Section 6 3 or we should NumCols 4 and log2ratioCol 4 for Affymetrix Genotyping Console Section 6 2 We check data import entering gt dataAffy Object of class setupGADA Affy data Number of probes 1813441 0 missing values Number of probes by chromosome 1 2 3 4 5 6 7 8 9 10 11 141348 148812 123956 116379 112136 109149 97441 95116 79106 90328 86362 12 13 14 15 16 17 18 19 20 21 22 84371 64071 55219 51570 52002 44888 50461 29067 41816 24208 23172 X Y 84315 8148 We can also visualize the raw data in a plot like in Figure 16 using gt plotRatio dataAffy num points 50000 The same information can be detailed as in Fig
25. 67 RefSeq Genes RefSeq Genes i HH Mammalian Gene Collection Full ORF mRNAs 96836 HM 40501 HI BC030613 011656 BC065294 025401 041803 BC114219 BC030228 E 022041 030570 4 112118 4 127666 Hill BC130319 BC040501 BC064534 Hi 127667 Hil BC130321 BC041803 4 BC009710 036407 BC098376 H H BC030570 4 4762 BC037876 BC112116 4 BC111600 BC108690 033942 Human mRNAs from GenBank Human mRNAs m Hk HEHH HH HH AHHH HHHH HEUTE HHHH H Figure 21 Results on the UCSC browser depicting a known CNV region 44 8 Connection with Aroma Affymetrix can also be called from within Aroma Affymetrix package http groups google com group aroma affymetrix which provides a normalization model described in 1 as well as an analysis frame work which includes copy number detection and visualization In this environment GADA segmentation tools are coupled with Aroma Affymetrix package pipeline In the vignette in http groups google com group aroma affymetrix web total copy number analysis 6 0 is performed by using GadaModel instead of CbsModel Following the vignette we have the set of instructions library aroma affymetrix lt AffymetrixCdfFile fromChipType GenomeWideSNP 6 tags Full Specify library files cs lt AffymetrixCelS
26. 9 72578384 72581327 1 707572010 1 1 20 72581344 72582418 4 0 787632927 1 1 21 72583514 72583724 4 2 123788677 1 1 23 105820716 105825648 21 0 346537724 1 1 666 143439162 143445353 3 1 216180667 i 668 147536256 147554906 14 0 844484429 670 153177486 154582680 634 0 498188104 1 671 154616633 154887040 42 0 015083500 1 673 4613756 4665080 3 1 142382667 Y 1 675 5584359 5620349 4 1 075439750 Y 1 The visual representation of the recovered segments can be obtained by gt plotRatio step2 chr 17 37 7 2 Analyzing a collection of 90 Affymetrix arrays Affymetrix data for 60 CEU samples are available at http www creal cat jrgonzalez GADA 20081114 AffyGTC301 rar We begin downloading all the txt files exported GTC3 or APT in the local rawData folder No other other file should be placed here since all files ending with txt in that folder will be imported gt ParAffyData lt setupParGADAaffy log2ratioCol 4 NumCols 4 Creating objects of class setupGADA for all input files Applying setupGADAaffy for 90 samples Importing array 06985 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 1 done Importing array 06991 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 2 done Importing array 12892 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 90 done Creating objects of class setupGADA for all input files done Remember to modify lo
27. 900669 62923765 72541525 72570001 stop num 62900162 62922795 72541505 72569615 72582431 mean num 0 0434 1 1810 0 0314 2 4720 1 7178 count num 37489 5 6739 26 Extracting regions from all fits done gt print cnrs controli0 chromosome start stop mean count begin figure h begin center label fig log intensities includegraphics width 3in height linewidth angle 90 graficas log_intensities eps caption Illumina log2ratio intensities by chromosome end center end figure 1 1 51599 62900162 0 04341482 37489 1 62900669 62922795 1 18099960 5 3 1 62923765 72541505 0 03139820 6739 21 1 241190561 241195976 1 29550644 4 22 1 241201331 247191012 0 04825901 3903 Note that GADA two step approach is however not fully exploited within the aroma affymetrix framework If we want to adjust T and MinSegLen to a higher value we will obtain sparser results and reduce the FDR gt cnrs lt getRegions gada arrays 1 chromosomes 1 T 20 MinSegLen 30 Repeating BackwardElimination List of 2 T num 20 MinSegLen num 30 gt print cnrs control10 chromosome start stop mean count 1 1 51599 72541505 0 04171265 44233 1 72541525 72583737 2 30345358 45 3 1 72584492 247191012 0 04046404 102246 1 http genome ucsc edu cgi bin hgTracks clade vertebrate amp org Human amp db hg18 amp position chr1 2 http genome ucsc edu cgi bin hgTracks clade vertebrate amp org Human amp db hg18 amp po
28. 926 1510418607 2760 19 59549722 0 8394 0 4610 1 6296 0 8394 0 4416 1 7768 BB 2 2860 21 33649200 NC 0 0232 0 9055 0 8936 0 5011 0 9227 0 8765 BB 1296 19 35839708 BB 0 8567 0 9771 1 8410 88 0 8567 0 9648 1 7772 BB 1516 22 41070487 0 8010 0 0527 0 5638 0 1475 0 6260 0 8030 You can drag and drop columns to re arrange their display order Also you can 1447 19 48275526 NC 0 0000 0 6095 2 2190 NC 0 0000 0 6355 2 1541 NC draa them to from the Hidden list to hide show them 3060 6 116796071 0 9181 0 5268 1 1214 0 9268 0 5371 1 0936 1447 19 34416353 0 949 0 0252 1 2934 0 9496 0 0267 1 1967 Display Locked Columns p Hidden Columns 14 X 97843727 0 6681 0 0228 0 8666 0 6681 0 9636 0 6119 0 Manifest 1426 22 43150427 0 9211 0 0463 0 6450 0 9211 0 0446 0 6455 14 al GenTrain Score 1 3160 22 36009709 0 9003 0 0361 1 0920 0 9003 0 0338 0 9175 Frac 1296 19 9256584 NC 0 0000 0 4682 0 8809 BB 0 8963 0 9801 0 6786 BB 1296 20 45994629 BB 0 8457 0 9451 1 0375 0 8512 0 9553 1 0287 BB 2 Hide gt FracG 3260 19 36530218 BB 0 9509 0 9868 1 2238 0 9509 0 5409 1 6458 Displayed Frac T 3360 19 281116 BB 0 8826 0 9684 1 3955 0 8826 0 9839 1 5017 Index Show 1447 19 34648448
29. 965 1 51674 0 0436751 CN_473981 1 52771 0 40294 CN_473982 1 52788 0 134605 497981 1 62627 0 0063667 CN 502615 1 75787 0 677508 CN 502613 1 75849 0 343111 31 Select the columns to export Allele Difference CNState Log2Ratio LOH SmoothSignal Figure 14 Exporting log2ratios from the Affymetrix genotyping console 3 0 GTC3 CN_502614 1 76175 1 44067 6 3 Exporting data from Affymetrix power tools Alternativelly the log2ratio intensities can be extracted with the Affymetrix power tools APT available from http www affymetrix compartners programs programs developer tools powertools affx This tools provide more flexibility on the normalization procedures and settings that can be used apt copynumber workflow adapter type normalization true reference output results dir MySamplesReference ab ref set analysis name MySamples cdf file GenomeWideSNP 6 cdf chrX probes GenomeWideSNP 6 chrXprobes chrY probes GenomeWideSNP 6 chrYprobes special snps GenomeWideSNP 6 specialSNPs netaffx snp annotation file GenomeWideSNP_6 na25 annot csv netaffx cn annotation file GenomeWideSNP 6 cn na25 annot csv delete files true o results dir text output true delete files false cel files CEL 32 Copy Number LOH Analysis Options Sample Type Unpaired Select Analysis Configuration To add a new Analysis Configuration please Cancel and select Copy Numbe
30. Backward Elimination with T 6 and minimum length size 3 sigma2 0 0266 chromosome discontinuities 1 1 55 2 2 69 3 3 28 4 4 50 5 5 37 6 6 27 7 7 45 8 8 36 9 9 13 10 10 23 11 11 27 12 12 20 13 13 25 14 14 21 15 15 22 16 16 16 17 17 18 18 18 14 19 19 6 20 20 16 21 21 5 22 22 15 23 X 60 24 Y 4 The advantage of using a two step approach is that we can flexibly adjust T remove or add breakpoints that will follow in significance without fitting the entire SBL model again gt summary step2 Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 6 and minimum length size 3 Number of segments 676 Base Amplitude of copy number 2 chr 1 22 0 0024 X 0 5423 Y 0 5282 Gains 1 and Loses 1 with respect Base Amplitude 36 Chromosome 17 0 5 1 0 EHET Figure 18 Affymetrix log ratio intensities and segments for chromosome 17 IniProbe EndProbe LenProbe MeanAmp chromosome State 1 51586 76192 10 0 408413323 1 1 3 1617766 1640775 11 1 104301596 1 1 4 1642243 1662451 7 1 967756180 1 f 6 5146486 5147215 3 0 696282657 1 f 7 5150357 17063437 6590 0 014075678 1 i 8 17076072 17131709 49 0 191480187 1 1 10 25468522 25519264 21 0 259745800 1 1 12 40794663 40800797 4 0 629174323 1 1 14 72528689 72541492 11 0 474053142 1 i 15 72541512 72547710 8 2 256061427 1 1 16 72551656 72569602 17 1 231081324 1 1 17 72569988 72575080 4 1 812702177 1 1 1
31. Mean 3rd Qu Max 618 7434 27930 114200 101700 16320000 Number of Total Segments by chromosome segments Gains Losses Chromosome 1 397 87 310 Chromosome 2 305 64 241 Chromosome 3 328 111 217 Chromosome 4 304 78 226 Chromosome 5 144 45 99 Chromosome 6 191 57 134 Chromosome 7 208 66 142 Chromosome 8 207 15 192 Chromosome 9 70 4 66 Chromosome 10 73 22 51 Chromosome 11 88 8 80 Chromosome 12 150 30 120 Chromosome 13 46 4 42 Chromosome 14 156 41 115 Chromosome 15 125 30 95 Chromosome 16 85 21 64 Chromosome 17 156 73 83 Chromosome 18 43 5 38 Chromosome 19 46 11 35 Chromosome 20 75 51 24 40 Chromosome 6 90 samples 2 1 2 75 5 50 09 x 1 4 0 Li I li 1 1 1 1 1 1 e 1 1 I 1 J i 1 1 1 1 1 1 1 1 1 I 1 ial 2 NE 5 2 f 2 5 1 1 1 1 I 1 1 1 1 Ei 1 b 1 1 1 1 I 1 1 1 i 1 1 1 THEM Sif BE m EEHEN rz NEN T T EN DUD 559595 5 9555244 a 220 200 2 2022222222 RRR RR BR 215024441 2 222 222 RVR DDR 888 N M Ra NN S NNER ERRERA Y bw See 2 i ho i56 b gash o Sho Sh Figure 19 Gains red and losses blue for 90 CEU individuals o
32. T 4 5 and minimum length size 3 sigma2 0 0141 chromosome discontinuities 1 1 36 2 2 30 3 3 39 4 4 22 5 5 28 6 6 26 7 7 29 8 8 33 9 9 24 10 10 16 11 11 41 12 12 23 13 13 17 14 14 17 15 15 11 16 16 9 17 17 8 18 18 16 19 19 3 20 20 10 21 21 6 22 22 4 23 X 43 24 Y 0 The SBL function returns the number of discontinuities for each chromosome the number of iterations and the tolerance given to the SBL algorithm to converge The BackwardElimination function gives the number of segments for each chromosome adjusted by the parameter T and the minimum number of consecutive altered probes in the argument MinSegLen We would like to highlight the advantage of using a two step approach We can flexibly adjust T remove or add breakpoints that will follow in significance without having to fit the entire SBL model again As T and MinSegLen increase the number of CNA breakpoints decreases Finally the altered segments defined between the modeled breakpoints are reported using the summary method We classify the segments gain and losses using a simple threshold on the segment mean amplitude i e MeanAmp gt summary step2 Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 4 5 and minimun length size 3 Number of segments Base Amplitude of copy number 2 chr 1 22 0 0174 X 0 221 Y 0 0275 Gains 1 and Losses 1 with respect Base Amplitude EndProbe LenProbe 496 498 500 504 506
33. and the genomic positions of each probe stored in the gen info variable described above Both segments and gen info for the HapMap samples are readily available at http groups google com group gadaproject From these quantities a reduced matrix is obtained with the command 21 mat f lt getReducedData segments gen info varSimil 0 99 subVariation 0 90 cnv blocks lt attr mat f cnv blocks matrixPlot mat f Matrix reduction is controlled with the parameters varSimil and subVariation Blocks of neighboring probes that do not differ in more than varSimil across all the subjects in all populations are created as new summary variables Their values are those of the first probes in each block detailed in the attribute attr mat f cnv blocks A further reduction is performed with subVariation In this case blocks that are relatively constant across subjects are discarded as uninformative Specifically if subVariation is set to 0 9 then only columns taking different values for at least 10 across all samples are kept The final matrix concatenates the populations into subject blocks and can be displayed by matrixPlot Note that the columns of the matrix are factors consistent with the fact that copy number status losses no change and gains are categorical variables Their levels are 1 0 1 The resulting matrix of the example can be downloaded into the working directory at http groups google com group gadaproject and recovered into t
34. are clearly differentiated with the first two principal axes that account for 2796 of the total variance 23 5 Association analysis Association analysis for multiple CNVs can be performed using multiCNVassoc function We will use a simulated data for which case control information is generated for HapMap samples We have randomly generated cases and controls In order to find some signals we have generated different proportion of cases for population Data is available at the google group page http groups google com group gadaproject web testing and can be loaded by typing load HapMap270reducedData Rdata First we perform a MCA in order to consider population stratification We save the first two eigen vectors to be used in the adjusted models dd lt dudi acm fortran mat f scan FALSE nf 5 comp1 lt dd 1il 1 comp2 dd 1i 3 After that association crude analysis is done by typing 0 lt multiCNVassoc mat f casco 1 We can also consider population stratification in the association analysis by fitting the following models lt multiCNVassoc mat f casco compi comp2 A list containing the CNVs associated with cases sorted by p values corrected for multiple comparisons using Benjamini Hockberg s method is obtained as following gt getPvalBH ansO cnv blocks 1 20 region pval pval BH 1 BlkCnv42Chri7 7 063842e 19 1 040504e 15 2 BlkCnv43Chri7 1 716604e 17 1 264279e 14 3 BlkCnv44Chri7 2 896770e
35. ction of Affymetrix array data The function setupParGADAaffy should be used in the case of having data from Affymetrix The perfor mance of this function is similar to the previous one gt myExampleAffy lt setupParGADAaffy log2ratioCol 4 NumCols 4 Creating objects of class setupGADA for all input files Applying setupGADAaffy for 90 samples Importing array 06985 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 1 done Importing array 06991 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 2 done Importing array 12892 GW6 C CN5 CNCHP myAffyData txt Read 7253768 items Array 90 done Applying setupGADAaffy completed succesfully 3 2 Segmentation procedure Once raw data is imported as objects of class setupGADA we can perform segmentation procedure for all individuals one by one The procedure for analyzing Illumina and Affymetrix data is the same Here we use the example of Illumina data to illustrate how to perform parallel segmentation procedure The Appendix shows an example for Affymetrix data To perform segmentation procedure for multiple arrays we use the function parSBL that repeatedly calls the function SBL The syntax is similar to those used in the function SBL gt parSBL myExample estim sigma2 TRUE aAlpha 0 8 Creating SBL directory done Retrieving annotation data done Segmentation procedure for 8 samples Array 1 The estimated sigma2
36. ed setting Sort FALSE will reduce computing time Not run datalllumina2 setupGADAIllumina file dataIllumina txt log2ratioCol 6 NumCols 6 sort FALSE End not run Other arguments such as saveGenInfo or orderProbes are used internally and it is not recommended to change them 2 2 Summarizing data Imported data is summarized entering the name of the object of class setupGADA or using the generic method print Object of class setupGADA Illumina data Number of probes 561303 118 missing values Number of probes by chromosome 1 2 3 4 5 6 7 8 9 10 11 12 13 42075 45432 37768 33705 34649 36689 30170 31880 26874 29242 27272 27143 20914 log ratio 14 15 16 17 18 19 20 21 22 X Y 18429 16625 16870 14341 16897 9501 14269 8251 8462 13835 10 Figure 2 2 shows the log ratio intensities and obtained with plotRatio This function has several ar guments that can be used to draw different types of plots This visualization tools require the package plotrix which is available from CRAN It can be installed by typing install packages plotrix By default plotRatio produces an output like Figure 2 2 which displays the entire genome gt plotRatio datalllumina Loading required package plotrix Intensities along the karyotype of a single chromosome e g chr 12 Figure 2 2 are displayed with gt plotRatio datalllumina chr 12 13 14 15 16 17 18 1920 8 1 Chromosome Figure 1 Illumina log2
37. ename Copy Number LOH Results Group Remove Copy Number LOH Results Group NAQ7022_GW6_C CNS 07022 GW6 C canary v1 SNP_A 2029077 41 602 401 Log2Ratio 0 66 Region CNP2269 Call 4 Conf 0 942720 Status Messages 11 15 2008 3 32 59 PM Copy number data exported to H DataSets Affy6 0G TC2output 2008111 4 batch1 NA12814_GW6_C CN5 CNCHP mpAffyDal 11 15 2008 3 Copy number data exported to H DataSets Affy6 0G T C2output 20081114 batch1 N412815_GW6_C CN5 CNCHP mpAffyDal 11 15 Copy number data exported to H DataSets Affy6 0GT C2output 20081 11 4 batchT NA12864_GW6_C CN5 CNCHP myAffyD al 117157 i number data exported to H DataSets Affy6 0G T C2output 20081 11 4 batch1 NA12865_G W6_C CN5 CNCHP myAffyDal J nat ES HEADS ac ALTONA ATA b atah ALAA C86 0 ORE OMAIN Library path H AfFymetrix GeneChip 4ffy_Data Library User Profile Roger Figure 13 Exporting log2ratios from the Affymetrix genotyping console 3 0 GTC3 Only the log2ratio intensities are necessary to be exported Figure 14 Once exported the resulting files containing the data can be found on the output folder that was specified on the Copy Number Tool Figure 14 The resulting files for each file should have the following format comments comments comments ProbeSet Chromosome Position Log2Ratio CN_473963 1 51586 0 257667 CN_473964 1 51659 0 264712 CN_473
38. ep R Parent2 4424530410_ 01 01 Samples File ep ri ep Index 1 4424530410 80102 Estimated Fle 5 3 MB Hee Se 4424530410 RO2COI 1 eal Size le 4424530410 RO2CO2 lt 2 Rows 48 Disp 48 Sel 0 Filer Filer is not active Rows 0 Disp 0 5 0 Filer Filer is not active 30 Figure 12 Exporting log2ratios from BeadStudio tool in different files 6 2 Exporting data from Affymetrix genotyping console GTC The Affymetrix Genotyping Console 3 0 GTC3 which can be downloaded from http www affymetrix com products services software can also be used to extract normalized log2ratio intensities from a collection of CEL files After analyzing the data with the GTC3 copy number tool the raw instensities are exported by selecting Export Copy Number LOH Results with the right button Figure 13 Affymetrix Genotyping Console CN Heat Map Workspace Data Sets HapMap ale amp amp Q iw sc l Reee GenomeWideSNP_S E Sample Attributes B tensity Data CNP2270 1 HA In Bounds li Out of Bounds Genotype Results D 20081114_144119 8 15 Number LOH Results E Show Copy Number QC Summary Table cannary2 Show Copy Number Segments L Reports E SNP Lists Show Copy Number Custom Regions Run Segment Reporting Tool 2 View Results in Browser View Results in Heat Map Export Copy Number LOH Results Le R
39. er Filer is not active Figure 10 Exporting log2ratios from BeadStudio tool in a unique file 28 BeadStudio Genotyping WGG0014 Jia Edit Analysis Window o SNP Graph X Full Data Table SNP Table Paired Sample Table 1 x mW rer z im m B i 710419363 amp 220 B Allele Log B Allele Log g Bal L Name Chr Postion Ratio 6796 Freq Ratio STP Freq Ratio 6796 rn ie c 7510319302 19 38874483 0 4613 0 0526 0 4856 0 0140 AB 0 4587 0 0083 2 Report 0 0023 0 1177 0 0080 0 1251 0 0 B 0 9827 0 4606 1 0000 0 4915 1 0 Genotyping Report 0 4391 0 0321 BB 0 9618 0 1643 0 4 i 0 0221 0 1776 0 0290 0 4491 0 0 What type of report would you like to generate B 0 9556 0 1512 89702 0 1772 09 Up 0 5528 0 1059 0 5339 0 0780 0 0 0 4060 0 4918 0 8714 0 0665 NC 0 4 5 0 0030 0 1881 0 0000 0 1136 0 0 2 om B 0 9969 1 1695 0 9741 1 1553 0 9 Final Report 0 0017 0 6188 AB 0 5974 0 1900 05 B 0 9875 0 8038 1 0000 0 7338 04 0 0060 0 2526 0 0057 0 0934
40. et fromName MeduloWithControls cdf cdf Defining folder with CEL files acc lt AllelicCrosstalkCalibration cs Set allelic crosstalk model csC lt process acc verbose verbose Fit and correct allelic crosstalk plm lt AvgCnPlm csC mergeStrands TRUE combineAlleles TRUE shift 300 Summarization model fit plm verbose verbose Fit summarization model ces lt getChipEffectSet plm fln lt FragmentLengthNormalization ces PCR fragment length normalization FLN V cesN lt process fln verbose verbose Once the normalization model is normalized and calibrated the following function adds the GADAmode1 methods to Aroma Affymetrix gt library gada gt addGadaToAromaAffymetrix Adds the gadaModel to aroma affymetrix We can create a GADAmodel using gt lt GadaModel cesN aAlpha 0 8 T 6 MinSegLen 3 Without reference gt print gada GadaModel Name MeduloWithControls Tags ACC ra XY AVG 300 A B FLN XY a0 8 Chip type virtual GenomeWideSNP_6 Path gadaData MeduloWithControls ACC ra XY AVG 300 A B FLN XY a0 8 GenomeWideSNP_6 Number of chip types 1 Chip effect set amp reference file pairs Chip type 1 of 1 GenomeWideSNP_6 Chip effect set CnChipEffectSet Name MeduloWithControls Tags ACC ra XY AVG 300 A B FLN XY Path plmData MeduloWithControls ACC ra XY AVG 300 A B FLN XY GenomeWideSNP_6 Platform Affymetrix Chip type GenomeW
41. g2ratioCol and NumCols if another format is used Once we have imported the data we can follow exactly the same steps as in the Illumina case in Section 3 We can store the object with the path structure for future analysis to avoid importing data again gt Storing object with the imported data save ParAffyData file ParAffyData rData load ParAffyData rData M Individual arrays and chromosomes are easily accessed for visualization plot ratio intensities for sample 4 plotRatio ParAffyData Sample 4 num points 5000 plot ratio intensities for sample 4 and chromosome 2 plotRatio ParAffyData Sample 4 chr 2 num points 5000 V VM MM M The segmetnation analysis can be run as a batch for all samples or in parallel if we have the snow and Rmpi packages installed gt Segmentation for all samples gt parSBL ParAffyData aAlpha 0 5 estim sigma2 TRUE Creating SBL directory done Retrieving annotation data done Segmentation procedure for 90 samples Array 1 The estimated sigma2 0 02662487 Array 1 done Array 2 The estimated sigma2 0 03188311 Array 2 done Array 90 The estimated sigma2 0 02679275 Array 90 done Segmentation procedure for 90 samples done Warning messages 1 In FUN 1 24 24L SBL algorithm did not converge after 10000 iterations and change 3 82616197214247e 06 2 In FUN 1 24 24L SBL algorithm did not converge after
42. gada package User s manual Roger Pique Regi Alejandro Caceres and Juan R Gonzalez April 7 2010 Abstract The gada package is the implementation of a flexible and efficient analysis pipeline to detect genomic copy number alterations from quantitative data The package can import the raw copy number normal ized intensities provided by Illumina BeadStudio Affymetrix powertools or any similar format Probes of different samples are split into separate files and can be analyzed on a standalone workstation or in parallel using a cluster multicore computer The speed and accuracy of the genome alteration detection analysis GADA approach combined with parallel computing results in one of the fastest and most accurate methods available GADA is especially suitable to extract copy number alterations CNAs on genomewide studies that utilize high density arrays of millions of markers to sample hundreds of subjects Contents 1 Installation 2 2 Analysis of a single array 2 2 1 Importing and preparing array data the setupGADA class 2 2 1 1 Creating a setupGADA object using setupGADAgeneral 2 2 1 2 Creating a setupGADA object for Illumina or Affymetrix array 3 2 2 Summarnzimng data 4 9 ar Geis ox OR LE e Gee ER OR oe 4 2 3 Copy number segmentation with SBL and BackwardElimination 7 3 Multiple array analysis 12 Sel dated Se Sondeo as fe ee
43. he R session by load HapMap270reducedData Rdata from which the multivariate analysis of the example can be resumed We use a discrimination analysis based on multiple correspondence analysis to rank variables ac cording to the variable s correlation to the axis defined by each population in the principal component subspace The discrimination and ranking of the variables follows from cp lt discrimin cnv mat f pop cla var rank lt rank variables cp cnv blocks cnv blocks where pop cla is population labeling for each subject The entries of this vector and rows of mat f must correspond to the same subject The first rows of the the ranking for the HapMap sample illustrate the most relevant blocks of variables in the discrimination of particular populations gt var rank 1 10 probe num pr pos inf pos sup chr correlation population 1 BlkCnv6Chr3 1 11 46819191 46822621 3 0 816 YRI 2 BlkCnv3Chr3 1 17 46777922 46794735 3 0 808 3 BlkCnv42Chri7 1 92 41570665 41707908 17 0 764 4 BlkCnv43Chri7 1 3 41708649 41711411 17 0 756 BlkCnv44Chri7 1 4 41717787 41719992 17 0 748 6 BlkCnv2Chr3 1 5 46776822 46777250 3 0 741 YRI 7 BlkCnv50Chr4 1 5 69079062 69096368 4 0 710 CBH JPT 8 BlkCnv5Chr3 1 1 46807086 46807086 3 0 708 YRI 9 BlkCnv49Chr4 1 10 69057944 69079058 4 0 693 CBH JPT 10 BlkCnv48Chr4 1 1 69057756 69057756 4 0 693 CBH JPT The variable B1kCnv6Chr3 1 for instance stands for the binary variable i
44. he package enforces a strict directory structure on the working directory to perform the analysis of multiple samples However the only required directory to be set up by the user is that containing the raw data This is an example of how folders are organized after analysis is completed exampleBeadStudio txt rawData sample C1 sample C2 sample C3 sample C4 sample C5 sample C6 sample C7 sample C8 allSegments gen info Rdata genomicInfo sbli 8512 8513 5014 515 5016 8517 8518 1 segments2 segments3 segments4 segments5 segments6 segments7 segments8 setupGADA1 setupGADA2 setupGADA3 setupGADA4 setupGADA5 setupGADA6 setupGADA7 setupGADA8 3 1 Raw data The rawData directory must contain all data files corresponding to each individual from a particular assay Each data file must be organized as described in Section 2 1 3 1 1 Importing a collection of Illumina array data The user may have all information in a unique file as indicated in Section 6 1 In this case the user can obtain individual files from gada by using splitDataBeadStudio function To illustrate how to split the file into subject files we use the following files available at gt download file http www crea
45. ideSNP 6 Full monocell Number of arrays 66 Names control10 controli1 N813 Time period 2008 11 06 20 37 47 2008 11 06 20 37 56 Total file size 1778 65MB RAM 0 11MB Parameters probeModel chr pm mergeStrands logi TRUE combineAlleles logi TRUE Reference file average across arrays RAM 0 00MB or using a paired reference set gt lt GadaModel cesi cesReference aAlpha 0 8 T 6 MinSegLen 3 With reference gt print gada 45 the parameters aAlpha and MinSegLen control the settings of SBL BackwardElimination methods as we described in this manual To fit the model we enter gt fit 1 3 5 chromosomes c 1 17 22 verbose verbose or for the entire set of samples and chromosomes gt fit verbose verbose The results of the segmentation can be displayed using the graphical reporting tools implemented in aroma affymetrix package gt ceGada lt ChromosomeExplorer gada gt print ceGada ChromosomeExplorer Name MeduloWithControls Tags ACC ra XY AVG 300 A B FLN XY a0 8 Number of arrays 66 Path reports MeduloWithControls ACC ra XY AVG 300 A B FLN XY a0 8 GenomeWideSNP_6 gada RAM 0 00 gt process ceGada chromosomes c 19 22 23 verbose verbose gt display ceGada The Firefox 2 0 or newer is required to visualize this results Figure 22 ACC 2Cra 2C X YNS2 CAVGS202B30052CANS2BBS2C FLNS2C X Y 2C a
46. l cat jrgonzalez GADA exampleBeadStudio txt exampleBeadStudio txt 12 Notice that the three first columns of this file must contain annotation data This information is required and it includes the name of probe the chromosome and the genomic position The other columns correspond to each individual Note however that the information can be variable depending on the information we have obtained from BeadStudio In this example we saved the log2ratio and the B allele frequency Name Chr Position 1 Log Ratio C1 Allele Freq C2 Log Ratio C2 Allele Freq Log Ratio Allele Freq C4 Log Ratio C4 Allele Freq C5 Log Ratio C5 Allele Freq C6 Log R Ratio C6 B Allele Freq C7 Log R Ratio C7 B Allele Freq C8 Log R Ratio C8 B Allele Freq rs758676 7 12878632 0 1134 0 5215 0 0312 1 0000 0 0098 1 0000 0 0442 1 0000 0 1815 0 9942 0 1144 0 4990 0 5641 0 0000 0 0488 1 0000 rs3916934 13 103143536 0 2099 0 0014 0 1669 0 5361 0 2143 0 0062 0 0371 0 9955 0 3281 0 0048 0 2505 0 5249 0 2122 0 5427 0 0262 0 9912 rs2711935 4 38838852 0 0443 0 0000 0 0972 0 5094 0 1467 0 0109 0 1192 0 4951 0 0490 0 4725 0 0704 0 0001 0 1707 0 4987 0 2628 0 5125 rs17126880 1 64922104 0 0659 0 9888 0 0917 1 0000 0 0008 0 9986 0 0442 0 9959 0 0766 1 0000 0 0343 0 9949 0 0454 0 9989 0 0398 0 9930 rs12831433 12 4995220 0 0072 0 0043 0 0782 0 0006 0 0927 0 5063 0 2230 0 5282 0 0317 0 0026 0 2575 0 9929 0 1471 0 0000 0 0219 0 0066 gt s
47. library gada 2 Analysis of a single array 2 1 Importing and preparing array data the setupGADA class The first step for the analysis is to prepare a setupGADA object that encapsulates the array hybridization intensities and other information such as the marker position in the genome and the genotype in case of SNP markers If data is already loaded on R the object is obtained with the function setupGADAgeneral Otherwise data can be loaded with setupGADAIllumina from the text files exported by Illumina BeadStudio or with setupGADAaffy for files obtained with Affymetrix Genotyping Console Section 2 1 2 These functions can also be used with other array platforms of similar output format 2 1 1 Creating a setupGADA object using setupGADAgeneral If data is already available in then setupGADAgeneral directly creates setupGADA object like in the example Simulated data set seed 123456 cn lt rep c rep 1 1E5 100 rep 1 100 rep 1 1E5 4 Underlying copy number arrayData lt rnorm length cn mean log2 cn 1 sd 1 Simulated array dataSim setupGADAgeneral arrayData Prepared setupGADA object dataSim Object of class setupGADA log ratio data VVVVV Number of probes 800000 0 missing values Number of probes by chromosome No genetic information available where we use a simulated sample Annotation data if available can also be added through the argument gen info as a data frame The followi
48. n chromosome 6 Chromosome 21 10 3 7 Chromosome 22 101 40 61 gt An increase of T will reduce the sensitivity to detect true breakpoints although the false discovery rate FDR will also be smaller We can plot the information that summarizes all the CNA findings using the functions plot and plotWG Figure 5 shows gains and losses across the whole genome while Figure 6 details the findings for chromosome 6 gt plotWG allSamples gt plot allSamples 6 show ind TRUE The probes that fall on areas containing CNA on chromosome 17 can also be obtained using gt altProbes getAlteredProbes allSamples chr 17 gt altProbes gains probe Freq chr pos 6 CN_146278 36 17 41523026 7 CN_146297 62 17 41581088 41 Gains Losses CNV frequency summary 90 samples 1 Sete ee ee ee eS 2 ee eee SS 1 Pg 5 re EE 6 e 7 SSS 3 a E e IT 0 1 o 514 Genomic Position 247190999
49. na and Affymetrix platforms to gada 6 1 Exporting data from Bead Studio The BeadStudio tool which is available at http www illumina com allows information to be provided in either a unique text file or a file per individual or groups of individuals The raw intensities for all individuals in a unique file can be exported shown in Figure 9 and Figure 10 Figure 9 shows how to select the columns of interest and Figure 10 how this information can be exported In order to obtain a different file for each individual the user has to select Final Report as illustrated in Figure 11 Then he must select 1 in the file Samples File Figure 12 Only Log R Ratio is necessary to be displayed but other fields such as B Allele Freq can be exported to analyze the data using other programs The resulting files for each subject have the following format if the user have selected genotype log2ratio and B allele frequency Name Chr Position GType Log R Ratio B Allele Freq rs10000010 4 21227772 BB 0 1157656 0 9982474 rs10000023 4 95952929 AB 0 1266638 0 4817977 rs10000030 4 103593179 AB 0 0514016 0 5103833 rs1000007 2 237416793 AB 0 138847 0 4689891 rs10000092 4 21504615 AA 0 01165604 0 00370151 rs10000121 4 157793485 BB 0 02247738 0 9751745 rs1000014 16 24325037 BB 0 0001281789 0 9989412 rs10000141 4 33810744 BB 0 01945104 0 9805357 rs1000016 2 235355721 AA 0 2437027 0 00094727 rs10000169 4 77575270 AA 0 08803905 0 rs100
50. ndicating deletions 1 in the CNV block number 6 in chromosome 3 The data frame encodes the corresponding number of segments in the block and its genomic inferior and superior positions As an example we show the percentage copy number alterations across populations for this particular block CBH JPT CEU YRI 0 100 97 21 1 0 3 79 Note that deletion in this block is indeed specific to the YRI population Selecting the first 87 variables which have correlations higher than 0 5 we finally perform multiple correspondence analysis on them as an unsupervised classification select lt getNamesProbes var proj min correlation 0 5 cm dudi acm mat f select scan FALSE nf 3 plot cm pop cla which axes c 1 2 var FALSE pnt 0 7 Figure 7 shows the classification that it is achieved with only 87 blocks of CNVs for the HapMap sample 22 re es A e d e e e cu 2 eo of i e e 4 44 IX e re tis LA gt ater ee S s m e e je e 2 I m lt 4 eu 1 le e e e e t foe i Me la to a o 2 l CBH JPT e CEU YRI t 1 0 0 5 0 0 0 5 Axis1 Figure 7 Multiple correspondence analysis for 87 CNV blocks of the HapMap sample Populations
51. ng format is required probe chr pos 1 rsi2354060 1 10004 2 rs6650104 1 554340 3 rs12184279 1 707348 4 rsi12564807 1 724325 5 rs3115860 1 743268 6 rs7515489 1 758845 7 1517160939 1 773886 8 rs12086311 1 798632 9 rs4475691 1 836671 10 rs28705211 1 890368 gt gen info lt data frame probe paste id 1 length cn sep chr c rep 1 2E5 rep 2 2E5 rep 3 2E5 rep 4 2E5 pos rep 1 length cn 4 4 10 gt setupGADA object with annotation information gt dataSim lt setupGADAgeneral arrayData gen info gen info gt dataSim Object of class setupGADA log ratio data Number of probes 800000 0 missing values Number of probes by chromosome 1 2 3 4 200000 200000 200000 200000 2 1 2 Creating a setupGADA object for Illumina or Affymetrix array Data exported from Illumina using BeadStudio tool or Affymetrix using Affymetrix genotyping console GTC or Affymetrix power tools APT are loaded using setupGADAIllumina or setupGADAAffy O functions respectively Sections 6 1 and 6 2 illustrates how to export data from both technologies In either case data must be arranged in the following format e 1st column probe e 2nd column chromosome e 3rd column genomic position e 4th column other information e e jth column log2ratio e kth column other information Two example files are provided for further detail The first one corresponds to an Illumina data example
52. of Total Segments segments Gains Losses 7913 2305 29 1 5608 70 9 Summary of length of segments Min ist Qu Median Mean 3rd Qu Max 533 5783 15050 67630 48990 21410000 Number of Total Segments by chromosome segments Gains Losses Chromosome 939 287 652 Chromosome 716 199 517 Chromosome 557 184 373 Chromosome 666 148 518 Chromosome 399 89 310 474 159 315 482 100 382 Chromosome 203 43 160 Chromosome 10 225 77 148 Chromosome 1 2 3 4 Chromosome 5 401 79 322 6 7 Chromosome 8 9 39 Chromosome 11 268 84 204 Chromosome 12 334 90 244 Chromosome 13 216 52 164 Chromosome 14 466 151 315 Chromosome 15 304 67 237 Chromosome 16 213 59 154 Chromosome 17 259 121 138 Chromosome 18 148 16 132 Chromosome 19 192 81 111 Chromosome 20 147 98 49 Chromosome 21 30 10 20 Chromosome 22 254 111 143 we can adjust the number of breakpoints by modifying the parameter very quickly we increase to 12 for example the number of detected breakpoints is reduced maintaining only those that are more likely to be true breakpoints gt parBE ParAffyData T 12 MinSegLen 10 gt allSamples lt summary ParAffyData length c 500 6e9 gt print allSamples NOTE 384 segments with length not in the range 500 6e 09 bases and with mean log2ratio in the range 0 28 0 16 have been discarded Number of Total Segments segments Gains Losses 3308 866 26 2 2442 73 8 Summary of length of segments Min ist Qu Median
53. ore ype 4424530410 RO2COI Theta 1 4424530410 80202 R Rows 48 Disp 48__ Sel 0__ Fiter Fiter is not active Rows 0_Disp 0_ Sel 0_Filter Filter is not active Figure 9 Exporting log2ratios from BeadStudio tool in a unique file 27 BeadStudio Genotyping WGG0014 gt Ele Analysis Tools Window d o SNP Graph 1 b x Full Data Table SNP Table Paired Sample Table 111 182 5 amp 2 1 200r 180r 160r 140r 120r iE E 2 osp 060 020 0 420r 207 E Guardar como 2 Guardar en C3 JR Samples Table Documentos recientes 4424530408_RO2CO1 Gg 4424530408 02 02 Escritorio 4424530409 RD1COL 4424530409 RD1COZ 7 Fiter Fiter is not active 4424530409 RO2COI 4424530409 02 02 amp 15x 4424530009 RO1COI 4424530009 01002 9 4424530009 RD2COI 4424530009 _ 02 02 Parent arent 4424530410_RO1CO1 ys index 4424530410 01 02 Index ES 4424530410 02 01 l 4424530410 RD2CO2 Nombre FllData Table Mis sitios dered Tipo Tab Delimited tt v Cancelar lt 2 Rows 48 Disp 48 Sel 0 Fiter Fiter is not active Rows 0 Disp 0 5 0 _Fit
54. plitDataBeadStudio exampleBeadStudio txt Samples 8 NumCols 5 Splitting data from BeadStudio btaining Ratio Intensity files NOTE individual files will be written to 8 files with name as indicated in header of input file btaining Ratio Intensity files done splitDataBeadStudio has two parameters Samples and NumCols The first one indicates the number of individuals analyzed in our case 8 samples The argument NumCols gives the number of columns considering the three first columns that contain the annotation data As we have information about log2ratio and B allele frequency the argument NumCols is set equal to 5 Once individual files are available we then import a collection of Illumina array data with gt myExample setupParGADAIllumina log2ratioCol 4 NumCols 5 Creating object with annotation data Read 3218460 items Creating object with annotation data done Creating objects of class setupGADA for all input files Applying setupGADAIllumina for 8 samples Importing array sample C1 Read 5364100 items Array 1 done Importing array sample C2 Read 5364100 items Array 2 done Importing array sample C3 Read 5364100 items Array 3 done Importing array sample C4 Read 5364100 items Array 4 done Importing array sample C5 Read 5364100 items Array 5 done Importing array sample C6 Read 5364100 items Array 6 done Importing array sample C7 Read 53641
55. r LOH Configurations from the Edit menu 1 Regional GC Correction Select Reference Model File Annotation File With SNP Markers Used For Reference Model File Annotation File With CN Markers Used For Reference Model File Select Output Root Path H DataS ets Affy6 0G T C2output Select CN LOH Batch Name 20081117 batch1 Dutput File Suffix leave blank for no suffix Figure 15 Copy number tool dialog box that specifies the results folder The APT user manual provide more detailed explanation on all the different possible settings We use text output option to produce the text files with the following GADA format head 500 NA06985_GW6_C MyTest CN5 CNCHP txt Comments Comments Comments ProbeSetName Chromosome Position CNState Log2Ratio SmoothSignal LOH Allele Difference CN_473963 1 51586 2 0 257667 1 054558 nan nan CN 473964 1 51659 2 0 264712 1 054389 nan nan CN 473965 1 51674 2 0 043675 1 054354 nan nan CN 473981 1 52771 2 0 402939 1 051817 nan nan CN 473982 1 52788 2 0 134605 1 051777 CN_497981 1 62627 2 0 006367 1 029375 nan nan CN 502615 1 75787 2 0 677508 1 000571 nan nan CN 502613 T 75849 2 0 343111 1 000438 nan nan CN 502614 1 76175 0 1 440673 0 999744 CN_502616 1 76192 0 2 477916 0 999708 CN_502843 1 88453 2 0 135097 0 974336 nan nan CN 466171 1 218557 2 0 030157 1 597003 nan nan CN 468414 1 218926 2 0 475484 1 597018 nan nan CN
56. ratio intensities by chromosome All plots that we have previously illustrated can be saved as a encapsulated postscript eps file using the postscript R function gt postscript file log intensities eps gt plotRatio datalllumina postscript TRUE gt dev off Visualizing all the probes on a single plot may generate an unnecessarily big file given the high resolu tion of the array platforms We can reduce the number of points in the plot modifying the argument num points gt postscript file log intensities eps gt plotRatio dataIllumina postscript TRUE num points 50000 gt dev off Chromosome 12 06 04 02 00 02 045 06 Figure 2 Illumina log2ratio intensities for chromosome 12 2 3 Copy number segmentation with SBL and BackwardElimination The segmentation procedure is divided in two steps as described in 2 The first step fits a sparse Bayesian learning SBL model and finds the most likely candidate breakpoints for the copy number state The second step implements a backward elimination BE procedure to remove sequentially the least significant breakpoints estimated by the SBL model allowing a flexible adjustment of the False Discovery Rate FDR The first step is implemented on the SBL procedure gt stepi SBL datalllumina estim sigma2 TRUE The estimated sigma2 0 01411465 and requires the setupGADA object e g dataIllumina in previous
57. section The SBL is controled by two parameters 1 the array noise level o and 2 the sparseness hyperparameter The array noise level c is estimated automatically by the algorithm by setting estim sigma2 TRUE If o is known a priori it be set manually with the assignment sigma2 c The sparseness hyperparameter aa aAlpha controls the SBL prior distribution which is uninformative about the location an amplitude of the CNA breakpoints but imposes a penalty on the number of CNA breakpoints A higher aa implies that less breakpoints are expected a priori and results with fewer true detected yet fewer false positives An efficient adjustment of the trade off between sensitivity and FDR is performed with a backward elimination BE procedure using a high sensitivity of ag 0 2 set up by default aAlpha 0 2 second step a backward elimination procedure implemented in BackwardElimination is used to adjust the FDR Step2 BackwardElimination stepi T 4 5 MinSegLen 3 where the T argument is the critical value of the BE algorithm for the statistical score tm associated to brake point m That is tm lower than T are discarded The score tm is the difference between the sample averages of the probes falling on the left and right segment divided by a pooled estimation of the standard error Asymptotically when the number of probes on the right and left segments are very large the distribution of score will converge
58. sition chr1 3 http genome ucsc edu cgi bin hgTracks clade vertebrate amp org Human amp db hg18 amp position chr1 However if we want to visualize the results with a higher value of T we have to repeat the entire procedure gt gada lt GadaModel cesN aAlpha 0 8 T 20 MinSegLen 30 gt process ceGada chromosomes c 19 22 23 force TRUE verbose verbose 47 In a future versions will consider reusing the previously fitted SBL model to only repeat the backward elimination step 48 References 1 H Bengtsson R Irizarry B Carvalho and T P Speed Estimation and assessment of raw copy numbers at the single locus level Bioinformatics 24 6 759 767 2008 2 R Pique Regi J Monso Varona A Ortega R C Seeger T J Triche and S Asgharzadeh Sparse representation and bayesian detection of genome copy number alterations from microarray data Bioinformatics 24 3 309 18 2008 toLatex sessionInfo e R version 2 7 0 2008 04 22 1686 pc linux gnu e Locale LC_CTYPE es_ES UTF 8 LC_NUMERIC C LC_TIME es_ES UTF 8 LC_COLLATE es_ES UTF 8 LC MONETARY C LC I e Base packages base datasets graphics grDevices methods stats utils Other packages gada 0 7 4 Loaded via a namespace and not attached tools 2 7 0 49
59. to a standard normal distribution i e A 0 1 The argument MinSegLen can be used to limit the minimum number of probes each CNA segment must contain We recommend using MinSegLen 3 default to eliminate false detections due to extreme outliers The following settings a and T are recommended depending on the desired sensitivity and FDR higher sensitivity higher FDR gt 0 2 gt 3 lt gt as 0 5 T gt 4 lower sensitivity lower FDR lt gt aa 0 8 T gt 5 The print generic function gives the user the following information for each step gt stepi Sparse Bayesian Learning SBL algorithm Sigma2 0 0141 chromosome discontinuities numit tolerance 1 1 1811 1434 9 884051 09 2 2 2013 917 9 955136e 09 3 3 1653 660 9 958175e 09 4 4 1558 2521 9 791328e 09 5 5 1551 581 9 990603e 09 6 6 1671 1968 9 951486e 09 7 T 1373 762 9 854208e 09 8 8 1330 1252 9 893025e 09 9 9 1148 900 9 599827e 09 10 10 1233 791 9 993864e 09 11 11 1165 1374 9 927208 09 12 12 1084 645 9 877164 09 13 13 976 697 9 869698 09 14 14 816 705 9 919774e 09 15 15 661 423 9 806253e 09 16 16 643 571 6 404507e 09 17 17 478 3302 9 226409e 09 18 18 751 468 9 694729e 09 19 19 383 849 9 933759e 09 20 20 521 1362 9 955395e 09 21 21 374 716 9 933086e 09 22 22 359 440 9 927054e 09 23 X 1472 1926 9 949394e 09 24 Y 1 50 8 340744e 09 gt step2 Sparse Bayesian Learning SBL algorithm SBL and Backward Elimination with
60. ts 260 Base Amplitude of copy number 2 chr 1 22 0 0989 X 0 3088 Y 3 4627 Array FO cee eo oa a ea oe Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 8 and minimun length size 8 Number of segments 100 Base Amplitude of copy number 2 chr 1 22 0 0159 X 0 1274 Y 0 0553 Backward elimination procedure for 8 samples done The function stores the segments in the directory SBL The arguments are the same as those used in the function BackwardElimination previously described 3 2 1 Paralell segmentation We have programmed setupParGADAIllumina setupParGADAaffy parSBL and parBE functions to allow the user to parallelize in few steps the analysis when multiple processors are available This has been implemented using the snow package After loading the required packages snow and Rmpi gt library snow gt library we create the cluster c1 We use the instruction gt cl makeCluster 8 type MPI 8 slaves are spawned successfully O failed Further examples including how to connect more than one workstation can be found in http www sfu ca sblay R snow htn gada library is loaded in all processors with gt clusterEvalQ cl library gada No further setting up is required After this when calling parSBL the computing time will automat ically decrease depending on the number of processors connected to the cluster 3 8 Summaryzing results We use the generic function s
61. ummary to collect all segments for each individual in single object gt allSamples lt summary myExample Warning message In summary parGADA myExample All segments are reported If you want to filter the minimum and maximum length of segments adjust length base e g length base c 500 10e6 in base units This function returns and object of class sumnaryParGADA The warning message is used to alert the user that all segments will be reported In some situations one is only interested in segments with a given size do so the parameter length base should be changed as we illustrate later Using the generic funtion print the following information is obtained 16 gt allSamples NOTE 814 segments with length not in the range O Inf bases and with mean log2ratio in the range 0 24 0 14 have been discarded Number of Total Segments segments Gains 7 Losses 444 38 8 6 406 91 4 Summary of length of segments Min 1st Qu Median Mean 3rd Qu Max 2169 20510 58600 221200 167700 8547000 Number of Total Segments by chromosome segments Gains Losses Chromosome 1 34 2 32 Chromosome 2 23 2 21 Chromosome 3 16 0 16 Chromosome 4 26 0 26 Chromosome 5 16 2 14 Chromosome 6 74 9 65 Chromosome 7 14 2 12 Chromosome 8 29 2 27 Chromosome 9 12 0 12 Chromosome 10 18 5 13 Chromosome 11 23 3 20 Chromosome 12 10 1 9 Chromosome 13 5 0 5 Chromosome 14 15 1 14 Chromosome 15 12 0 12 Chromosome 16 32 2 30 Chromosome 17 25 2 2
62. ure Number of segments 878 Base Amplitude of copy number Array 4 2 Sparse Bayesian Learning SBL Backward Elimination procedure Number of segments 151 Base Amplitude of copy number Array 3 Sparse Bayesian Learning SBL Backward Elimination procedure Number of segments 208 Base Amplitude of copy number Array 4 Sparse Bayesian Learning SBL Backward Elimination procedure Number of segments 542 Base Amplitude of copy number Array 5 Sparse Bayesian Learning SBL Backward Elimination procedure Number of segments 560 Base Amplitude of copy number Array 6 Sparse Bayesian Learning SBL Backward Elimination procedure Number of segments 82 done for 8 samples algorithm with T 8 and minimun length size 8 2 chr 1 22 0 0274 X 0 0433 Y 0 1369 algorithm with T 8 and minimun length size 8 2 chr 1 22 0 0097 X 0 0844 Y 0 0637 algorithm with T 8 and minimun length size 8 2 chr 1 22 0 0118 0 0377 1 0436 algorithm with T 8 and minimun length size 8 2 chr 1 22 0 0054 X 0 3908 Y 4 3155 algorithm with T 8 and minimun length size 8 2 chr 1 22 0 0056 X 0 3985 Y 4 1761 algorithm with T 8 and minimun length size 8 15 Base Amplitude of copy number 2 chr 1 22 0 0023 0 0708 Y 0 0727 Array ET eaa SES SSRs ea as ee SS ee eee Sparse Bayesian Learning SBL algorithm Backward Elimination procedure with T 8 and minimun length size 8 Number of segmen
63. ure 17 for chromosome 12 gt plotRatio dataAffy chr 12 num points 50000 The segments are obtained by the two step approach consisting of the SBL and BackwardElimination procedures as described in section 2 3 gt step1 lt SBL dataAffy aAlpha 0 5 estim sigma2 TRUE The estimated sigma2 0 02658385 gt step1 Sparse Bayesian Learning SBL algorithm sigma2 0 0266 chromosome discontinuities numit tolerance 1 1 3011 1087 9 941786e 09 2 2 3168 897 9 971203 09 3 3 2439 690 9 890185e 09 4 4 2458 2630 9 973668e 09 5 5 2213 1697 9 998423e 09 6 6 2134 591 9 958981e 09 7 7 2080 1270 9 936370e 09 8 8 1913 1489 9 976727e 09 9 9 1651 3488 9 969362e 09 10 10 1926 734 8 913446e 09 34 X 0 24 4 d one1 60 Chromosome Figure 16 Affymetrix log ratio intensities by chromosome Chromosome 17 0 5 4 1 0 15 Figure 17 Affymetrix log ratio intensities for chromosome 17 35 11 11 1700 661 9 812538 09 12 12 1733 520 9 690517 09 13 13 1385 564 9 858486 09 14 14 1271 682 9 96073 09 15 15 1069 433 9 733707e 09 16 16 1156 618 9 905047 09 17 17 1026 457 9 924282e 09 18 18 991 680 9 499992 09 19 19 720 327 6 731085 09 20 20 843 413 9 827422e 09 21 21 555 360 9 616085 09 22 22 600 371 9 603308 09 23 X 4016 2775 9 506493e 09 24 Y 183 608 3 030268e 09 gt step2 BackwardElimination stepi T 6 MinSegLen 3 gt step2 Sparse Bayesian Learning SBL algorithm SBL and
Download Pdf Manuals
Related Search