@prefix LoadListingRequirement: .
@prefix CommandLineBinding: .
@prefix xsd: .
@prefix DockerRequirement: .
@prefix Workflow: .
@prefix sld: .
@prefix rdfs: .
@prefix CommandOutputBinding: .
@prefix ns1: .
@prefix NetworkAccess: .
@prefix rdf: .
@prefix cwl: .
@prefix ResourceRequirement: .
a cwl:CommandLineTool ;
rdfs:label "bgzip VCF" ;
ns1:original_cwlVersion "v1.2" ;
cwl:arguments ( "-c" ) ;
cwl:arguments ( "-c" ) ;
cwl:baseCommand ( "/usr/local/bin/bgzip" ) ;
cwl:baseCommand ( "/usr/local/bin/bgzip" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/samtools:1.11--h6270b1f_0"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/samtools:1.11--h6270b1f_0"
] ;
cwl:stdout "$(inputs.file.basename).gz" .
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.vcf.basename)" ] ;
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.vcf.basename)" ] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
4 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
-3 ] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
-4 ] ;
sld:type cwl:File .
sld:type cwl:File .
cwl:in ;
cwl:out ;
cwl:run .
cwl:source .
cwl:in ;
cwl:out ;
cwl:run .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"docm_filtered_variants.vcf" ] ;
sld:type cwl:File .
cwl:in , , , ;
cwl:out ;
cwl:run .
a cwl:CommandLineTool ;
rdfs:label "Filter variants from the DoCM detector" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "$(runtime.outdir)/docm_filtered_variants.vcf" ) ;
cwl:baseCommand ( "/usr/bin/perl" "docm_filter.pl" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/cle:v1.4.2"
] ;
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. Provide docm_vcf, normal_cram, tumor_cram, output_vcf_file, set_filter_flag\") unless scalar(@ARGV) == 5;\nmy ($docm_vcf, $normal_cram, $tumor_cram, $output_vcf_file, $set_filter_flag) = @ARGV;\n\nmy $samtools = '/opt/samtools/bin/samtools';\nmy $normal_header_str = `$samtools view -H $normal_cram | grep \"^\\@RG\" | head -n 1`;\nmy $tumor_header_str = `$samtools view -H $tumor_cram | grep \"^\\@RG\" | head -n 1`;\n\nmy ($normal_name) = $normal_header_str =~ /\\@RG.+\\tSM:([ -~]+)/;\nmy ($tumor_name) = $tumor_header_str =~ /\\@RG.+\\tSM:([ -~]+)/;\n\nunless ($normal_name and $tumor_name) {\n die \"Failed to get normal_name: $normal_name from $normal_cram AND tumor_name: $tumor_name from $tumor_cram\";\n}\n\nmy $docm_vcf_fh;\nif($docm_vcf =~ /.gz$/){\n open($docm_vcf_fh, \"gunzip -c $docm_vcf |\") or die(\"couldn't open $docm_vcf to read\");\n} else {\n open($docm_vcf_fh, $docm_vcf) or die(\"couldn't open $docm_vcf to read\");\n}\nopen(my $docm_out_fh, \">\", \"$output_vcf_file\") or die(\"couldn't open $output_vcf_file for write\");\n\nmy ($normal_index, $tumor_index);\n\nwhile (<$docm_vcf_fh>) {\n chomp;\n if (/^##/) {\n say $docm_out_fh $_;\n }\n elsif (/^#CHROM/) {\n if ($set_filter_flag) {\n say $docm_out_fh '##FILTER=';\n }\n my @columns = split /\\t/, $_;\n my %index = (\n $columns[9] => 9,\n $columns[10] => 10,\n );\n ($normal_index, $tumor_index) = map{$index{$_}}($normal_name, $tumor_name);\n unless ($normal_index and $tumor_index) {\n die \"Failed to get normal_index: $normal_index for $normal_name AND tumor_index: $tumor_index for $tumor_name\";\n }\n $columns[9] = $normal_name;\n $columns[10] = $tumor_name;\n my $header = join \"\\t\", @columns;\n say $docm_out_fh $header;\n }\n else {\n my @columns = split /\\t/, $_;\n my @tumor_info = split /:/, $columns[$tumor_index];\n my ($AD, $DP) = ($tumor_info[1], $tumor_info[2]);\n next unless $AD;\n my @AD = split /,/, $AD;\n shift @AD; #the first one is ref count\n \n for my $ad (@AD) {\n if ($ad > 5 and $ad/$DP > 0.01) {\n my ($normal_col, $tumor_col) = map{$columns[$_]}($normal_index, $tumor_index);\n $columns[9] = $normal_col;\n $columns[10] = $tumor_col;\n if ($set_filter_flag) {\n $columns[6] = 'DOCM_ONLY';\n }\n else {\n $columns[6] = '.';\n }\n my $new_line = join \"\\t\", @columns;\n say $docm_out_fh $new_line;\n last;\n }\n }\n }\n}\n\nclose($docm_vcf_fh);\nclose($docm_out_fh);\n" ;
cwl:entryname "docm_filter.pl"
]
] ;
cwl:requirements [ a cwl:StepInputExpressionRequirement ] .
cwl:source .
cwl:source .
cwl:in ;
cwl:out ;
cwl:run .
cwl:in ;
cwl:out ;
cwl:run .
cwl:source .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
2 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , sld:null .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"docm_raw_variants.vcf" ] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position 1 ;
cwl:valueFrom "${\n if(inputs.filter_docm_variants){\n return \"1\";\n } else {\n return \"0\";\n }\n}\n"
] ;
sld:type xsd:boolean .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.file.basename).gz" ] ;
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.file.basename).gz" ] ;
sld:type cwl:File .
cwl:source .
cwl:source .
cwl:in ;
cwl:out ;
cwl:run .
cwl:source .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
3 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
sld:type xsd:boolean .
a cwl:CommandLineTool ;
rdfs:label "run vt decompose" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-s" "-o"
[ cwl:valueFrom "$(runtime.outdir)/decomposed.vcf.gz" ]
) ;
cwl:baseCommand ( "vt" "decompose" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/vt:0.57721--hf74b74d_1"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] .
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
a cwl:CommandLineTool ;
rdfs:label "HaplotypeCaller (GATK4)" ;
ns1:original_cwlVersion "v1.0" ;
cwl:baseCommand ( "/bin/bash" "docm_haplotypeCaller.sh" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "broadinstitute/gatk:4.1.2.0"
] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "set -o pipefail\nset -o errexit\n\n# Running haplotype caller using the newly created interval list\nif [[ \"$#\" == 5 ]];then # If normal_bam is passed.\n # explicitly capturing variables\n reference=$1\n normal_bam=$2\n tumor_bam=$3\n docm_vcf=$4\n interval_list=$5\n # Chaning the interval_list to a new docm_interval_list that spans the docm regions by 200bp\n cat $interval_list | grep '^@' > docm.interval_list # Extracting the header from the interval_list\n zcat $docm_vcf | grep ^chr | awk '{FS = \"\\t\";OFS = \"\\t\";print $1,$2-100,$2+100,\"+\",$1\"_\"$2-100\"_\"$2+100}' >> docm.interval_list # Extracting the docm regions with a 100bp flanking region on both directions\n /gatk/gatk HaplotypeCaller --java-options \"-Xmx8g\" -R $reference -I $normal_bam -I $tumor_bam --alleles $docm_vcf -L docm.interval_list --genotyping-mode GENOTYPE_GIVEN_ALLELES -O docm_raw_variants.vcf\nelse # If normal_bam is not passed\n reference=$1\n tumor_bam=$2\n docm_vcf=$3\n interval_list=$4\n # Chaning the interval_list to a new docm_interval_list that spans the docm regions by 200bp\n cat $interval_list | grep '^@' > docm.interval_list # Extracting the header from the interval_list\n zcat $docm_vcf | grep ^chr | awk '{FS = \"\\t\";OFS = \"\\t\";print $1,$2-100,$2+100,\"+\",$1\"_\"$2-100\"_\"$2+100}' >> docm.interval_list # Extracting the docm regions with a 100bp flanking region on both directions\n /gatk/gatk HaplotypeCaller --java-options \"-Xmx8g\" -R $reference -I $tumor_bam --alleles $docm_vcf -L docm.interval_list --genotyping-mode GENOTYPE_GIVEN_ALLELES -O docm_raw_variants.vcf\nfi\n" ;
cwl:entryname "docm_haplotypeCaller.sh"
]
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 9000
] .
a cwl:Workflow ;
rdfs:label "Detect Docm variants" ;
ns1:original_cwlVersion "v1.0" ;
Workflow:steps , , , , , , ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] .
cwl:inputBinding [ CommandLineBinding:position
5 ] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type cwl:File .
cwl:outputSource ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position 1 ;
cwl:valueFrom "$(self.basename)"
] ;
cwl:inputBinding [ CommandLineBinding:position 1 ;
cwl:valueFrom "$(self.basename)"
] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
-2 ] ;
sld:type cwl:File .
cwl:in , , , , ;
cwl:out ;
cwl:run .
cwl:source .
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:outputBinding [ CommandOutputBinding:glob
"decomposed.vcf.gz" ] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
a cwl:CommandLineTool ;
rdfs:label "vcf index" ;
ns1:original_cwlVersion "v1.2" ;
cwl:arguments ( "cp" "$(inputs.vcf.path)" "$(runtime.outdir)/$(inputs.vcf.basename)"
[ CommandLineBinding:shellQuote false ;
cwl:valueFrom " && "
]
"/usr/local/bin/tabix" "-p" "vcf"
) ;
cwl:arguments ( "cp" "$(inputs.vcf.path)" "$(runtime.outdir)/$(inputs.vcf.basename)"
[ CommandLineBinding:shellQuote false ;
cwl:valueFrom " && "
]
"/usr/local/bin/tabix" "-p" "vcf"
) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:ShellCommandRequirement ] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/samtools:1.11--h6270b1f_0"
] ;
cwl:requirements [ a cwl:StepInputExpressionRequirement ] ;
cwl:requirements [ a cwl:ShellCommandRequirement ] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/samtools:1.11--h6270b1f_0"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:StepInputExpressionRequirement ] .