# HG changeset patch -- Bitbucket.org # Project galaxy-dist # URL http://bitbucket.org/galaxy/galaxy-dist/overview # User jeremy goecks <jeremy.goecks@emory.edu> # Date 1288902634 14400 # Node ID cdd6ce1f38b0e22802ed63e3e0cb4f6a58cb93b2 # Parent f372ea5a601443a9b9cd1d3f7e04ddc24db5b57a Enable filter tool to operate correctly on files with a variable number of columns (such as SAM files). Tool prepares and processes only columns up to and including the right-most column (i.e. column with the largest index) in the filtering condition, and subsequent columns are ignored. New functional test added for this functionality as well. --- a/tools/stats/filtering.py +++ b/tools/stats/filtering.py @@ -60,10 +60,18 @@ for operand in operands: except: if operand in secured: stop_err( "Illegal value '%s' in condition '%s'" % ( operand, cond_text ) ) + +# Find the largest column used in the filter. +largest_col_index = -1 +for match in re.finditer( 'c(\d)+', cond_text ): + col_index = int( match.group()[1:] ) + if col_index > largest_col_index: + largest_col_index = col_index -# Prepare the column variable names and wrappers for column data types +# Prepare the column variable names and wrappers for column data types. Only +# prepare columns up to largest column in condition. cols, type_casts = [], [] -for col in range( 1, in_columns + 1 ): +for col in range( 1, largest_col_index + 1 ): col_name = "c%d" % col cols.append( col_name ) col_type = in_column_types[ col - 1 ] @@ -72,7 +80,7 @@ for col in range( 1, in_columns + 1 ): col_str = ', '.join( cols ) # 'c1, c2, c3, c4' type_cast_str = ', '.join( type_casts ) # 'str(c1), int(c2), int(c3), str(c4)' -assign = "%s = line.split( '\\t' )" % col_str +assign = "%s, = line.split( '\\t' )[:%i]" % ( col_str, largest_col_index ) wrap = "%s = %s" % ( col_str, type_cast_str ) skipped_lines = 0 first_invalid_line = 0 --- a/tools/stats/filtering.xml +++ b/tools/stats/filtering.xml @@ -23,6 +23,13 @@ <param name="cond" value="c1=='chr1' and c3-c2>=2000 and c6=='+'"/><output name="out_file1" file="filter1_test2.bed"/></test> + <!-- Test filtering of file with a variable number of columns. --> + <test> + <param name="input" value="filter1_in3.sam"/> + <param name="cond" value="c3=='chr1' and c5>5"/> + <output name="out_file1" file="filter1_test3.sam"/> + </test> + </tests><help> --- /dev/null +++ b/test-data/filter1_in3.sam @@ -0,0 +1,100 @@ +HWI-EAS269B:8:34:797:623 145 chr1 16632182 255 27M = 16631977 0 CCATTTCCTGTATGCTGTAAAGTACAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:40:39:1184 145 chr1 24620331 3 27M = 24620115 0 ATTTATGTGGTTTCGTTTACCTTCTAT IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:2 CC:Z:chrM CP:i:9060 +HWI-EAS269B:8:58:533:1198 81 chr1 88426079 255 27M = 88423429 0 GAAGAGGAAGAAGGTGGGGAGGAAGAG IIIG?IIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:89:1776:1815 97 chr1 134085638 255 27M = 134085824 0 GAATGATTCTCTGGGTGTTACTTTGCA IIIIIIIIIIIIIIIDIII:IIII>F5 NM:i:0 NH:i:1 +HWI-EAS269B:8:74:1134:1670 161 chr1 138166886 255 27M = 138167084 0 TTACTAGTGTCTCTCTTACCATCATAT .IIIIIIIIIIIIIIIIIIIIIIIIII NM:i:1 NH:i:1 +HWI-EAS269:3:59:1321:1427 147 chr1 173149715 255 27M = 173149555 0 AAGGGCTAGGGTGACAGGCAGGGGACG -C<CID?IIIIIIIDIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:8:164:1678 145 chr1 178660716 255 27M = 178660493 0 CAATTGGTGTTTTTCTTAAGAGACTCA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:1:1048:638 137 chr10 12456478 1 27M * 0 0 TAAAANAAATAAAACAAAACAATAAAA $'&(+")'%$(&*&#&$$#%%$$%%$$ NM:i:2 NH:i:4 CC:Z:chr16 CP:i:21639623 +HWI-EAS269:3:62:1211:798 147 chr10 28135294 255 27M = 28135117 0 ACATGGTGTGGGGACAGAGATGTGAAG I;IIIIIIAIIIIIIIIIIIIIIIIII NM:i:1 NH:i:1 +HWI-EAS269:3:27:410:424 145 chr10 76950070 255 27M = 76949871 0 GAGTCTGTGTCCAGGCCAATTCACTAT 25-8I+.6B.IIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:48:1180:1122 73 chr10 83772157 0 27M * 0 0 GATCATCTTTTCTAAAACAATAAAGAC /IIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:6 CC:Z:chr11 CP:i:93811140 +HWI-EAS269:3:77:654:131 99 chr10 93419810 255 27M = 93419993 0 TTGCATCCCTAGGAACTGGAGTTATAG IIIIIIGIHIIIED@CIIH5I3D9G6: NM:i:0 NH:i:1 +HWI-EAS269B:8:98:895:1810 177 chr11 3371472 255 27M = 37952212 0 CTACATAGTGGGAACCGGCGACCGCTG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:2 NH:i:1 +HWI-EAS269B:8:58:1126:883 97 chr11 30080486 255 27M = 30080702 0 TACTTCCACCAGGCTCCAGTTTTGTGA IIIIIIIIIIIIIIIIIIIIIIIIII5 NM:i:0 NH:i:1 +HWI-EAS269B:8:89:1763:1446 81 chr11 59589605 255 27M = 59589400 0 GTACCTGGCCACTGATGCAGCTTAGAA II<IIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:6:739:1329 161 chr11 68629628 3 27M = 68629835 0 GTTTTTGGTTGATAGTTGAGCAAACTG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:2 CC:Z:chr4 CP:i:126661877 +HWI-EAS269:3:19:1683:1350 81 chr11 70846915 255 27M = 70846725 0 CACGGGCAGATAAGCTGCTGAGACTAA H50BB:IF=IIGIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:9:1691:1087 97 chr11 84640783 255 27M = 84641010 0 ACATTATCCATCTCTCTGTCATTGTCC IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:48:1350:424 81 chr11 94150331 255 27M = 94150113 0 TTGCAAGCCAACCCTGAGTGAAGTGTC IIDHIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:22:1176:1405 97 chr11 98629843 255 27M = 98630030 0 CCCTCGGAGGGCAACGGGAAAGGAGAA IIIIIIIIIIIDIIIII?III/79E14 NM:i:0 NH:i:1 +HWI-EAS269:3:67:1311:1512 97 chr11 100488190 255 27M = 100488374 0 CATGCCCTGAGACTTAGCAAGACTCTT IIIIIIIIIIIIIIIEIA3,I57GBI@ NM:i:0 NH:i:1 +HWI-EAS269B:8:41:1142:1307 161 chr11 101320424 255 27M = 101320625 0 GCCTCCCTACATAGCAAAGGAAAGAAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:31:1101:21 97 chr11 116711287 255 27M = 116711485 0 CATAAGCAAAAGATACACCATGTTTTA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:84:637:1203 145 chr12 35829424 255 27M = 35829229 0 CTCCCTCAAGGATCTTGTGGGGCATCA </'>I,I5B26$II;EB=%IIII1III NM:i:0 NH:i:1 +HWI-EAS269:3:65:1158:1081 163 chr12 52234332 255 27M = 52234505 0 TTTTTTTTTTAAGACAGGGAGTTTTTT IIIIIIIIIIIC4,,*>II%-,%III> NM:i:1 NH:i:1 +HWI-EAS269B:8:65:1325:1287 137 chr12 52234513 3 27M * 0 0 GCAGCACCTTGTGACATTTAATTTAGT IIIII+IIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:2 CC:Z:chr15 CP:i:98780443 +HWI-EAS269:3:38:488:1093 97 chr12 80471642 255 27M = 80471839 0 TGGAGAGATGACTCCTTGTTTGAGACA IIIIIIIIIIIIIIIIIIIBAE0B<7D NM:i:0 NH:i:1 +HWI-EAS269:3:76:290:1451 99 chr12 117745566 0 27M = 117745739 0 CAATCTGAAGATCCACAATCTTTTATA IIIIIIIIIIIIIIFII5I9IIII+I7 NM:i:0 NH:i:5 CC:Z:chr4 CP:i:133520935 +HWI-EAS269B:8:39:353:1526 65 chr13 18122558 255 27M = 18122761 0 ACTCTACTCAAAACCACACTAAGCCTC @IIIIII6IIIIIDII9IIIIII<III NM:i:0 NH:i:1 +HWI-EAS269:3:63:1260:365 99 chr13 41278147 255 27M = 41278319 0 AAGACAAGAACTTATCCACCAATATGT IIIIIIIIIIIIIII<IIA<CIIDII> NM:i:0 NH:i:1 +HWI-EAS269:3:70:152:1609 161 chr13 55421426 255 27M = 55421687 0 CGCTTGACCATTCCAGCCCAGACAAGA IIIIIIIII8IIII3IIII-C(3-%B' NM:i:0 NH:i:1 +HWI-EAS269:3:9:1343:1414 99 chr13 83721797 255 27M = 83721967 0 AACTACAGCTGAGGCAGCCTCCTGCCT IIIIIIIIIIIIIIIII;E+?7&819& NM:i:0 NH:i:1 +HWI-EAS269:3:4:1480:1956 137 chr14 46703849 0 27M * 0 0 GTGGAGCCCAGTGACACCATCGAGAAC IIIIIIIIIIIII?C3IH?1@I@@=27 NM:i:0 NH:i:8 CC:Z:= CP:i:46704077 +HWI-EAS269:3:86:616:1519 137 chr14 56244763 255 27M * 0 0 CATGGCCTGAAGTTCCTGAGCTTTATC IIIIIIIII3DIBI9II3)73BIG'G+ NM:i:0 NH:i:1 +HWI-EAS269:3:42:656:1216 97 chr14 100271567 255 27M = 100271751 0 ATAAGTCTCAGTTCTTGGGCCAGATCA IIIIIIIIICIIIIII3IA854I1C+) NM:i:0 NH:i:1 +HWI-EAS269:3:87:95:767 73 chr16 7368880 255 27M * 0 0 AATATCTGAAACATTCAAATGAGCAAT F>II/4I@9H=II?IIIIIIIIIIII- NM:i:0 NH:i:1 +HWI-EAS269B:8:50:1572:711 145 chr16 32230400 255 27M = 32230196 0 TACAGCAACAACAAATTCAACGACACG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:56:1436:121 147 chr16 49887529 255 27M = 49887356 0 TTAAGGCCCAGCTCTACATAAAACACT IIIIAHIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:53:584:1696 99 chr17 45705815 255 27M = 45705987 0 CCTTCTTGTCCAAGATCTCCTTCATGA IIIIIIIIIIIIBI6IIII=II:GI@< NM:i:0 NH:i:1 +HWI-EAS269B:8:16:643:1950 177 chr18 38000738 255 27M = 38000570 0 ATTCTGGTTCAGCCTGGGCAGCTTGGG III@==I0<IIGIII;IIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:66:1760:1539 137 chr18 38624681 3 27M * 0 0 TTTGGAACTTTTGAGAGGATCCCTAGC IIIIIIIIIIIIIII1II.IIIII1II NM:i:0 NH:i:2 +HWI-EAS269:3:2:1723:1277 163 chr19 16341705 255 27M = 16341870 0 ATGGGCTCGTCGCAGCTCAGCGGCTGG IIIIIIIIIIIIICI=8<&I7:F(+;2 NM:i:1 NH:i:1 +HWI-EAS269B:8:66:317:1676 97 chr19 47099316 255 27M = 47099535 0 ATCTGGTCAATCAACACCACCAGCAGA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:10:1428:1315 81 chr2 22444224 3 27M = 22444039 0 GGAATTGCGATAATTATAGTGGCTGAT ?6>CI>GIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:2 CC:Z:chrM CP:i:6245 +HWI-EAS269:3:72:152:785 97 chr2 22796881 255 27M = 22797074 0 TGAGTACTCCAGATAATCGTTACACAA IIIIIIII6IIIIIIII0IA@?,7752 NM:i:0 NH:i:1 +HWI-EAS269:3:79:321:1095 99 chr2 62083760 255 27M = 62083938 0 CAAAAATTGAGATATCAAAAAGCTCTT IIIIIIIII2IGIBIIC8:4IB1H4I; NM:i:0 NH:i:1 +HWI-EAS269:3:93:1529:881 97 chr2 129998362 255 27M = 129998827 0 CGCATGCCAGGAGGGGGCATGCCCATT IIIFIIIIFIIIIIFII?72I.,020* NM:i:0 NH:i:1 +HWI-EAS269B:8:20:813:376 113 chr2 130534278 255 27M = 130534454 0 TGTATTAGCAGGAGGTGGGGAGGCTGA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:89:1221:1082 83 chr2 151285071 0 27M = 151284944 0 ACCAGTGCACAGGTCTCCAGGGCTTCT -IC=B+8IIIEIIFI5IIIIIIIIIII NM:i:0 NH:i:5 CC:Z:chr9 CP:i:31743980 +HWI-EAS269:3:66:46:336 163 chr2 156679837 255 27M = 156680016 0 TATTTTCCTTTTGCTGTGGTTTGTGTT IIIIIIIIIIIIIEI@E-I?GI&B%3I NM:i:0 NH:i:1 +HWI-EAS269B:8:52:1139:1381 73 chr2 174953004 0 27M * 0 0 AATGCTCAACTCTTAGTTTCTTATTCA IIIIIIIIIIIIIIIIB@?IIIII?+0 NM:i:0 NH:i:19 CC:Z:= CP:i:175018018 +HWI-EAS269:3:98:585:19 161 chr3 26005087 255 27M = 26005271 0 ACCTAACACATGTAAACTTAAATTCAT I:IIIIIIII<IIE==DDI):,D1--8 NM:i:0 NH:i:1 +HWI-EAS269:3:74:447:1309 83 chr3 80492396 255 27M = 80492220 0 TCGGATGCCTCTCACCACTTTGACAAT )H():F=85IIIIEIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:74:425:1427 161 chr3 96447603 255 27M = 96447820 0 TGGAGTAGGAGTCTCAGGAGGAGTAGA IIIIIIIIIIIIIIIIIIIII:I?II< NM:i:0 NH:i:1 +HWI-EAS269:3:10:739:1474 163 chr3 124089371 255 27M = 124089535 0 AAAAAAACAATCTTATTCCGAGCATTC IIIIIIIIIIIIIIIIIIIIIIIGIII NM:i:0 NH:i:1 +HWI-EAS269:3:20:357:1709 65 chr4 128895987 255 27M = 128895822 0 CCTACCTTCTTCCCTTGGCAGCTGACT IIII>IIIBIIA:I18):,*3&,/*'+ NM:i:0 NH:i:1 +HWI-EAS269B:8:22:623:129 113 chr4 135492867 255 27M = 135490613 0 CCACTTTCCTGTACTGGCCAGAAAATG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:48:824:702 65 chr4 139584010 255 27M = 139582899 0 CTGGGCAGTGCAGCGGTACATGGAGCC IIIIIIIIIIIIIIIIII<AIIB;I>: NM:i:0 NH:i:1 +HWI-EAS269B:8:33:527:1211 73 chr5 37627410 255 27M * 0 0 GGGAAGGTGTGAGTACAACAGCCAAAG IIIIIIIIIIIIIIIIIIIIIIII@II NM:i:0 NH:i:1 +HWI-EAS269:3:2:1518:599 163 chr5 136908928 255 27M = 136909101 0 CTATTGCCAAAAAACTATGTTCACAAA IIIIIIIIIIIIIIIIIIIIIIIII=I NM:i:0 NH:i:1 +HWI-EAS269B:8:62:1493:1455 161 chr5 138715332 255 27M = 138715534 0 CCCAAATGAAAAAATAAATATTATGAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:69:384:1894 147 chr5 143665243 255 27M = 143665064 0 TGTTTGCTCCAACCAACTGCTGTCGCC &++*,2H-IGI+IFIGI?IIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:56:1358:762 113 chr6 15024174 255 7M12235N20M = 51485453 0 TGGGTACTTTCTCTAGCTCCTCCATTG /@)3I?IIIIIIIIIIIIIIIIIIIII NM:i:2 XS:A:- NH:i:1 +HWI-EAS269:3:100:228:1799 163 chr6 37756378 3 27M = 37756558 0 GTGTGGGGCTGCGTGGCCTGGCTGGTG DII4A+>5EI.)F634820&1(0%&&& NM:i:2 NH:i:2 +HWI-EAS269B:8:50:188:1253 161 chr6 52296684 1 27M = 52296901 0 ACTTTTCAGGGTTTTCAATAGTCACAC IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:1 NH:i:3 CC:Z:= CP:i:140695441 +HWI-EAS269B:8:19:440:1687 161 chr6 52850377 255 27M = 52850580 0 TACAAAGATGGACTTTTAAAATTCATT IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:47:1555:1520 161 chr6 67231655 255 27M = 67233919 0 ATTCAGTTATGGACCATCATTTCCGGA IIIIIIIIIIIIIIIIIIIIII<AII1 NM:i:0 NH:i:1 +HWI-EAS269B:8:79:1563:1318 161 chr6 89267414 255 27M = 89267614 0 GTGCGTGTTCCAGGCAGAGCTGGAAGA IIIIIIIIIIIIIIIIIIIIIII=;IA NM:i:0 NH:i:1 +HWI-EAS269B:8:54:954:565 161 chr6 120882513 255 27M = 120882716 0 ACGTCATGGCTGACCAGGACAGAGGTG IIIIIIIIIIIIIII?IEG;II<II*8 NM:i:0 NH:i:1 +HWI-EAS269B:8:16:570:1775 145 chr7 26081268 255 27M = 26080313 0 GAAAGAGTGACACAAATCAATAGTAAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:13:280:189 161 chr7 29184615 255 27M = 29184814 0 CTTTTCCAACAGCGAGAAAAATGTACA IIIIIIIIIIIIIIIIIIIIIIIIDII NM:i:0 NH:i:1 +HWI-EAS269B:8:19:72:273 137 chr7 52258521 255 27M * 0 0 ATCAGAGGCACAGGGACAGGGTAAGGA &&2IIIII;E(IIIIIIIIIIIIIIII NM:i:1 NH:i:1 +HWI-EAS269B:8:40:235:869 161 chr7 52631974 255 27M = 52632179 0 GGCGCTGACTCCATCAGATATCCATTC IIIIIII7IIIIIII6IIIIIIIBIII NM:i:1 NH:i:1 +HWI-EAS269B:8:18:304:346 129 chr7 107789557 255 27M = 33158390 0 CACGTACTGTCACCTTGTAACATTTGG IIIIIIIIIIIIIIIIIIIIIIIIIIF NM:i:0 NH:i:1 +HWI-EAS269B:8:3:698:297 161 chr7 110976065 255 27M = 110976385 0 GCAGTTATCACTTTCTTGCCATGGGCC IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:50:1269:965 145 chr7 118217605 255 27M = 118216239 0 ACCTGTAGATCCACATGATCATGAAGA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:38:1381:1429 83 chr7 136141344 255 27M = 136141280 0 ATCTGAAGTATCCCACATGTTGAGCTC <III@IIIIIIIII>IIIIIIIIIIII NM:i:1 NH:i:1 +HWI-EAS269:3:25:781:386 83 chr8 29844005 255 27M = 29843841 0 TAAGGGAGGAAAGTGTTTCAGAGTGTA ;83<159<<III@IIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:25:979:679 163 chr8 72664237 255 27M = 72664413 0 CTGATGGGAGCCCTGCGTGGTAAGAGG IIIIIIIIIII=III@I(II$27I>I4 NM:i:0 NH:i:1 +HWI-EAS269B:8:23:1069:816 145 chr8 74704433 255 27M = 74704222 0 TGTTCTCAGTTGTGGACAAGTGACAGC I<GII@IIIIIIII;IIIIIIIIII0I NM:i:0 NH:i:1 +HWI-EAS269B:8:34:435:1679 73 chr8 87366211 3 27M * 0 0 AAGCCTAGGGCTTCTCCTCTACACCCC I556I;FCAIIIFI<IIIIIIIIIIII NM:i:0 NH:i:2 +HWI-EAS269B:8:32:1486:294 97 chr8 124121215 255 27M = 124121411 0 TGTTACCATACGCCCTTCTGCTGAGGC IIIIIIIIIIIIIIIIIIDIIIIGIII NM:i:0 NH:i:1 +HWI-EAS269B:8:32:142:1471 81 chr8 124496752 3 27M = 124496536 0 GGATCTGCTTCATGAGTTGCCACATTG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:1 NH:i:2 +HWI-EAS269B:8:60:82:1806 97 chr8 125945215 255 27M = 125945423 0 AGATGCTGGCCATCCAGAGTAAGAACA IIIIIIIIIIIIIIIIIIIHIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:16:1162:495 81 chr8 125945584 255 27M = 125945381 0 GTTGCTCGCAGCTGGGGTGTGGGGCCA <CIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:1:165:1889 81 chr8 126016885 255 27M = 126016469 0 TGAGCAGGAAAACACTTTAAACCAGAT IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:66:939:1154 145 chr9 17537149 255 27M = 17536959 0 CTCCTTCAAGTACGCCTGGGTCTTAGA IDIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:68:1189:285 81 chr9 61249807 255 27M = 61242460 0 AGTCAAGCTCACTTGGCGGTGAAGGAT @.0;;A,?-F5I;7IIIIGIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:15:784:1754 163 chr9 74846937 255 27M = 74847108 0 GCACAGCACTGAGGAAAGGATCATCTC IIIIIIIIIIIIIIIIIIGII1CI7II NM:i:0 NH:i:1 +HWI-EAS269B:8:26:126:1382 145 chrM 3413 255 27M = 3190 0 GTTATTCTTTATAGCAGAGTACACTAA IIII8IB328I9IIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:90:320:1336 73 chrM 6326 255 27M * 0 0 ATGATCTCCAGCTATCCTATGAGCCTT IIIIIIIIA.ICI.4'(=,C>7-*&@8 NM:i:1 NH:i:1 +HWI-EAS269B:8:23:469:1215 161 chrM 7472 3 27M = 7676 0 ATTTCATCTGAAGACGTCCTCCACTCA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:2 +HWI-EAS269:3:73:36:783 161 chrM 7958 3 27M = 8152 0 CCCCAACAATAATAGGATTCCCAATCG IIIIIIIIIIIIIIIIIIIII09I>>I NM:i:0 NH:i:2 +HWI-EAS269:3:33:1528:954 99 chrM 8012 3 27M = 8186 0 TCCTATTCCCATCCTCAAAACGCCTAA IIIIIIIIIIIIIIIIIIIIIIIDII: NM:i:0 NH:i:2 +HWI-EAS269:3:2:192:1456 161 chrM 9071 3 27M = 9255 0 ACGAAACCACATAAATCAAGCCCTACT III;IEIIDI7III+III*?I@CH+5I NM:i:0 NH:i:2 +HWI-EAS269:3:63:192:470 97 chrM 14787 255 27M = 14982 0 GCAGATAAAATTCCATTTCACCCCTAC IIIIIIIIIIIIIIFIII=B5042:4E NM:i:0 NH:i:1 +HWI-EAS269B:8:49:528:63 83 chrM 14963 255 27M = 14829 0 TTTCCTATTTGCATACGCCATTCTACG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:98:856:1516 145 chrX 20257960 255 27M = 20256746 0 TACCCGGATTTAAGATGTACCCCATTG IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:28:1174:110 145 chrX 53916348 255 27M = 53916153 0 TGAATGTCAGCATCATTGACCCACAAA IIIIFIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:58:663:410 147 chrX 75006902 255 27M = 75006834 0 TCAGGTGGTTTACAGTGTTCTGACAAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 --- /dev/null +++ b/test-data/filter1_test3.sam @@ -0,0 +1,6 @@ +HWI-EAS269B:8:34:797:623 145 chr1 16632182 255 27M = 16631977 0 CCATTTCCTGTATGCTGTAAAGTACAA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:58:533:1198 81 chr1 88426079 255 27M = 88423429 0 GAAGAGGAAGAAGGTGGGGAGGAAGAG IIIG?IIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269:3:89:1776:1815 97 chr1 134085638 255 27M = 134085824 0 GAATGATTCTCTGGGTGTTACTTTGCA IIIIIIIIIIIIIIIDIII:IIII>F5 NM:i:0 NH:i:1 +HWI-EAS269B:8:74:1134:1670 161 chr1 138166886 255 27M = 138167084 0 TTACTAGTGTCTCTCTTACCATCATAT .IIIIIIIIIIIIIIIIIIIIIIIIII NM:i:1 NH:i:1 +HWI-EAS269:3:59:1321:1427 147 chr1 173149715 255 27M = 173149555 0 AAGGGCTAGGGTGACAGGCAGGGGACG -C<CID?IIIIIIIDIIIIIIIIIIII NM:i:0 NH:i:1 +HWI-EAS269B:8:8:164:1678 145 chr1 178660716 255 27M = 178660493 0 CAATTGGTGTTTTTCTTAAGAGACTCA IIIIIIIIIIIIIIIIIIIIIIIIIII NM:i:0 NH:i:1