@prefix rdf: . @prefix CommandLineBinding: . @prefix xsd: . @prefix DockerRequirement: . @prefix cwl: . @prefix Workflow: . @prefix ResourceRequirement: . @prefix sld: . @prefix rdfs: . @prefix CommandOutputBinding: . @prefix ns1: . cwl:in ; cwl:out ; cwl:run . cwl:outputSource ; cwl:secondaryFiles [] ; sld:type cwl:File . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.vcf.basename)" ] ; cwl:secondaryFiles [] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:label "add GT tags" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( "/usr/bin/perl" "add_strelka_gt.pl" "$(inputs.vcf.path)" "$(runtime.outdir)" ) ; cwl:cwlVersion ; cwl:hints [ a ] ; cwl:hints [ a ] ; cwl:inputs ; cwl:outputs ; cwl:requirements [ a cwl:InitialWorkDirRequirement ; cwl:listing [ cwl:entry "#!/usr/bin/perl\n\nuse strict;\nuse warnings;\n\nuse feature qw(say);\n\ndie(\"wrong number of arguments\") unless scalar(@ARGV) == 2;\nmy ($strelka_vcf, $outdir) = @ARGV;\n\nopen(my $strelka_vcf_fh, '-|', '/bin/gunzip', '-c', $strelka_vcf) \n or die(\"couldn't open $strelka_vcf to read\");\nopen(my $add_gt_fh, \">\", \"$outdir/add_gt.vcf\") \n or die(\"couldn't open add_gt.vcf for write\");\n\nwhile (<$strelka_vcf_fh>) {\n chomp;\n if (/^##/) {\n say $add_gt_fh $_;\n }\n elsif (/^#/) { #COLUMN HEADER\n say $add_gt_fh '##FORMAT=';\n say $add_gt_fh $_;\n }\n else {\n my @columns = split /\\t/, $_;\n my ($ref, $alt, $info) = map{$columns[$_]}(3, 4, 7);\n my @alts = split /,/, $alt;\n\n my ($n_gt, $t_gt);\n\n if (length($ref) == 1 and length($alts[0]) == 1) {\n my ($n_gt_info, $n_gt_str, $t_gt_str) = $info =~ /NT=(\\S+?);QSS.*SGT=(\\S+?)\\->(\\S+?);/;\n unshift @alts, $ref;\n\n my %ids;\n my $id = 0;\n\n for my $base (@alts) {\n $ids{$base} = $id;\n $id++;\n }\n\n $n_gt = $n_gt_info eq 'ref' ? '0/0' : parse_gt($n_gt_str, \\%ids);\n $t_gt = parse_gt($t_gt_str, \\%ids);\n }\n else {#INDEL\n my ($n_gt_info, $t_gt_info) = $info =~ /;NT=(\\S+?);.*SGT.*\\->(\\S+?);/;\n\n my %gt_info = (\n ref => '0/0',\n het => '0/1',\n hom => '1/1',\n conflict => './.',\n );\n\n $n_gt = $gt_info{$n_gt_info};\n $t_gt = $gt_info{$t_gt_info};\n }\n\n $columns[8] = 'GT:'.$columns[8];\n $columns[9] = $n_gt . ':' . $columns[9];\n $columns[10] = $t_gt . ':' . $columns[10];\n\n my $new_str = join \"\\t\", @columns;\n say $add_gt_fh $new_str;\n }\n}\n\nclose($strelka_vcf_fh);\nclose($add_gt_fh);\n\n\nsub parse_gt {\n my ($gt_str, $ids) = @_;\n my @gt_ids = map{$ids->{$_}}(split //, $gt_str);\n return join '/', sort @gt_ids;\n}\n\n#SNV example\n#1 10231 . C A . QSS_ref NT=ref;QSS=1;QSS_NT=1;SGT=AC->AC;SOMATIC;TQSS=2;TQSS_NT=2 DP:FDP:SDP:SUBDP:AU:CU:GU:TU 32:4:8:0:0,3:28,60:0,0:0,1 84:6:69:0:7,21:71,192:0,0:0,1\n#INDEL example\n##1 965051 . ATGTGTG A . QSI_ref IC=5;IHP=2;NT=ref;QSI=1;QSI_NT=1;RC=8;RU=TG;SGT=het->het;SOMATIC;TQSI=1;TQSI_NT=1 DP:DP2:TAR:TIR:TOR:DP50:FDP50:SUBDP50 8:8:6,6:0,0:2,4:10.3:0.00:0.00 18:18:8,8:5,6:5,8:21:0.25:0.00\n" ; cwl:entryname "add_strelka_gt.pl" ] ] ; cwl:requirements [ a cwl:ResourceRequirement ; ResourceRequirement:ramMin 4000 ] ; cwl:requirements [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "ubuntu:bionic" ] . cwl:outputBinding [ CommandOutputBinding:glob "$(inputs.file.basename).gz" ] ; sld:type cwl:File . a cwl:CommandLineTool ; rdfs:label "vcf index" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( "cp" "$(inputs.vcf.path)" "$(runtime.outdir)/$(inputs.vcf.basename)" [ CommandLineBinding:shellQuote false ; cwl:valueFrom " && " ] "/usr/bin/tabix" "-p" "vcf" ) ; cwl:cwlVersion ; cwl:hints [ a ] ; cwl:hints [ a ] ; cwl:inputs ; cwl:outputs ; cwl:requirements [ a cwl:ResourceRequirement ; ResourceRequirement:ramMin 4000 ] ; cwl:requirements [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgibio/samtools-cwl:1.0.0" ] ; cwl:requirements [ a cwl:ShellCommandRequirement ] . cwl:source . cwl:inputBinding [ CommandLineBinding:position 1 ; cwl:valueFrom "$(self.basename)" ] ; sld:type cwl:File . cwl:in ; cwl:out ; cwl:run . cwl:outputBinding [ CommandOutputBinding:glob "add_gt.vcf" ] ; sld:type cwl:File . sld:type cwl:File . cwl:inputBinding [ CommandLineBinding:position 1 ] ; sld:type cwl:File . cwl:source . cwl:source . a cwl:CommandLineTool ; rdfs:label "bgzip VCF" ; ns1:original_cwlVersion "v1.0" ; cwl:arguments ( "-c" ) ; cwl:baseCommand ( "/opt/htslib/bin/bgzip" ) ; cwl:cwlVersion ; cwl:hints [ a ] ; cwl:hints [ a ] ; cwl:inputs ; cwl:outputs ; cwl:requirements [ a cwl:DockerRequirement ; DockerRequirement:dockerPull "mgibio/samtools-cwl:1.0.0" ] ; cwl:requirements [ a cwl:ResourceRequirement ; ResourceRequirement:ramMin 4000 ] ; cwl:stdout "$(inputs.file.basename).gz" . cwl:in ; cwl:out ; cwl:run . a cwl:Workflow ; rdfs:label "process VCF workflow" ; ns1:original_cwlVersion "v1.0" ; Workflow:steps , , ; cwl:cwlVersion ; cwl:hints [ a ] ; cwl:hints [ a ] ; cwl:inputs ; cwl:outputs . sld:type cwl:File .