@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: .
rdfs:comment "Minimum number of sv calls needed to be merged" ;
cwl:inputBinding [ CommandLineBinding:position
3 ] ;
sld:type xsd:int .
rdfs:comment "assume genotypes at missing sites are 0/0" ;
cwl:default false ;
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "--missing-to-ref"
] ;
sld:type xsd:boolean , sld:null .
a cwl:Workflow ;
rdfs:label "Merge, annotate, and generate a TSV for SVs" ;
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:InlineJavascriptRequirement ] ;
cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] ;
cwl:requirements [ a cwl:StepInputExpressionRequirement ] ;
cwl:requirements [ a cwl:ScatterFeatureRequirement ] .
cwl:in ;
cwl:out ;
cwl:run .
cwl:in , , , ;
cwl:out ;
cwl:run .
cwl:in ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_vcf_name)" ] ;
sld:type cwl:File .
a cwl:CommandLineTool ;
rdfs:label "Run SURVIVOR to merge SV calls" ;
ns1:original_cwlVersion "v1.0" ;
cwl:baseCommand ( "/bin/bash" "/usr/bin/survivor_merge_helper.sh" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 1 ;
ResourceRequirement:ramMin 2000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/survivor-cwl:1.0.6.2"
] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] .
a cwl:CommandLineTool ;
rdfs:label "bgzip VCF" ;
ns1:original_cwlVersion "v1.2" ;
cwl:arguments ( "-c" ) ;
cwl:arguments ( "-c" ) ;
cwl:baseCommand ( "/opt/htslib/bin/bgzip" ) ;
cwl:baseCommand ( "/opt/htslib/bin/bgzip" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/samtools-cwl:1.0.0"
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/samtools-cwl:1.0.0"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:stdout "$(inputs.file.basename).gz" .
rdfs:comment "input bgzipped tabix indexed vcfs to merge" ;
cwl:inputBinding [ CommandLineBinding:position
6 ] ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "--ignore-pass-filter"
] ;
sld:type xsd:boolean , sld:null .
rdfs:comment "Maximum distance of variants to consider for merging" ;
cwl:inputBinding [ CommandLineBinding:position
2 ] ;
sld:type xsd:int .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.file.basename).gz" ] ;
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.file.basename).gz" ] ;
sld:type cwl:File .
cwl:default "v" .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.cohort_name)" ] ;
sld:type cwl:File .
cwl:default "bcftools-merged-AnnotSV.tsv" .
cwl:default "filtered-bcftools-merged-AnnotSV.tsv" ;
cwl:inputBinding [ CommandLineBinding:position 5 ;
CommandLineBinding:prefix "--output"
] ;
sld:type xsd:string , sld:null .
rdfs:comment "Require merged SVs to be on the same strand" ;
cwl:inputBinding [ CommandLineBinding:position 5 ;
cwl:valueFrom "${\n if(inputs.same_strand){\n return \"1\";\n } else {\n return \"0\";\n }\n}\n"
] ;
sld:type xsd:boolean .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "-genomeBuild"
] ;
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "-genomeBuild"
] ;
sld:type xsd:string .
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "--input"
] ;
sld:type cwl:File .
cwl:source .
sld:type xsd:int .
rdfs:comment "vcf file to filter" ;
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "-SVinputFile"
] ;
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "-SVinputFile"
] ;
sld:type cwl:File .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_tsv_name)" ] ;
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_tsv_name)" ] ;
sld:type cwl:File .
cwl:default "SURVIVOR-sv-merged.vcf" .
cwl:source .
rdfs:comment "output file format" ;
cwl:default "z" ;
cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "--output-type"
] ;
sld:type [ sld:symbols , , , ;
sld:type sld:enum
] .
cwl:default "0.05" .
rdfs:comment "Require merged SVs to be of the same type" ;
cwl:inputBinding [ CommandLineBinding:position 4 ;
cwl:valueFrom "${\n if(inputs.same_type){\n return \"1\";\n } else {\n return \"0\";\n }\n}\n"
] ;
sld:type xsd:boolean .
rdfs:comment "Array of VCFs to merge SV calls of" ;
cwl:inputBinding [ CommandLineBinding:itemSeparator
"," ;
CommandLineBinding:position 1
] ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
cwl:outputSource ;
sld:type cwl:File .
cwl:outputSource ;
sld:type cwl:File .
cwl:source .
sld:type xsd:int .
sld:type xsd:int .
a cwl:CommandLineTool ;
ns1:original_cwlVersion "v1.0" ;
cwl:baseCommand ( "/opt/bcftools/bin/bcftools" "merge" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/bcftools-cwl:1.9"
] .
sld:type cwl:File , sld:null .
cwl:source ;
cwl:valueFrom "${ return [ self ]; }" .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type cwl:File .
cwl:source ;
cwl:valueFrom "${ return [ self ]; }" .
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "--all-CDS"
] ;
sld:type xsd:boolean , sld:null .
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
sld:type xsd:boolean .
rdfs:comment "resolve duplicate sample names" ;
cwl:default true ;
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "--force-samples"
] ;
sld:type xsd:boolean , sld:null .
cwl:source .
cwl:source .
sld:type xsd:string .
cwl:default "bcftools-sv-merged.vcf" .
sld:type xsd:boolean .
cwl:source .
cwl:default "SURVIVOR-merged-AnnotSV.tsv" .
cwl:outputSource ;
sld:type cwl:File .
cwl:default "none" .
cwl:in , , , , , , , ;
cwl:out ;
cwl:run .
cwl:source .
a cwl:CommandLineTool ;
rdfs:label "filter AnnotSV tsv output generated from a bcftools merged vcf" ;
ns1:original_cwlVersion "v1.0" ;
cwl:baseCommand ( "python3" "AnnotSV_filter.py" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "import argparse\nimport csv\nimport sys\n\nparser = argparse.ArgumentParser()\nparser.add_argument('--input', '-i', dest=\"input\", help='input AnnotSV tsv file', required=True, action=\"store\")\nparser.add_argument('--output', '-o', dest=\"output\", help='output tsv file name', required=True, action=\"store\")\nparser.add_argument('--filtering_frequency', dest=\"filtering_frequency\", help=\"frequency to filter with\", action=\"store\", type=float, default=\"0.05\")\nparser.add_argument('--all-CDS', dest=\"CDS\", help=\"Do not require a positive CoDing Sequence overlap\", action=\"store_true\")\nparser.add_argument('--ignore-pass-filter', dest=\"filter\", help=\"Do not require calls to have a PASS filter\", action=\"store_true\")\n\nargs = parser.parse_args()\ninput_file_name = args.input\noutput_file_name = args.output\nfiltering_frequency = args.filtering_frequency\nall_cds = args.CDS\nignore_pass_filter = args.filter\n\nwith open(input_file_name, 'r') as file_in, open(output_file_name, 'w') as file_out:\n file_in = csv.DictReader(file_in, delimiter='\\t')\n file_out = csv.DictWriter(file_out, fieldnames=file_in.fieldnames, delimiter='\\t')\n file_out.writeheader()\n total_sv_count = 0\n pass_sv_count = 0\n for row in file_in:\n total_sv_count += 1\n if(row['AnnotSV type'] == 'split' \\\n and (row['FILTER'] == 'PASS' or ignore_pass_filter) \\\n and (int(row['CDS length']) > 0 or all_cds) \\\n and float(row['IMH_AF']) < filtering_frequency\n and float(row['1000g_max_AF']) < filtering_frequency\n and not(float(row['DGV_LOSS_Frequency']) > filtering_frequency and 'DEL' in row['SV type']) \n and not(float(row['DGV_GAIN_Frequency']) < filtering_frequency and ('DUP' in row['SV type'] or 'INS' in row['SV type']))\n and not(('Manta' in row['ID'] and 'IMPRECISE' in row['INFO']) or (row['QUAL'] != '.' and 'IMPRECISE' in row['INFO'])) ):\n file_out.writerow(row)\n pass_sv_count += 1\n print(\"total sv count:\",total_sv_count)\n print(\"total sv passed count:\",pass_sv_count)\n" ;
cwl:entryname "AnnotSV_filter.py"
]
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "python:3"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] .
cwl:source .
cwl:source .
cwl:outputSource ;
sld:type cwl:File .
sld:type xsd:string , sld:null .
cwl:source .
cwl:outputSource ;
sld:type cwl:File .
rdfs:comment "Estimate distance based on the size of SV" ;
cwl:inputBinding [ CommandLineBinding:position 6 ;
cwl:valueFrom "${\n if(inputs.estimate_sv_distance){\n return \"1\";\n } else {\n return \"0\";\n }\n}\n"
] ;
sld:type xsd:boolean .
cwl:source .
cwl:in , , , ;
cwl:out ;
cwl:run .
sld:type xsd:boolean .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_tsv_name)" ] ;
sld:type cwl:File .
cwl:source .
a cwl:CommandLineTool ;
ns1:original_cwlVersion "v1.2" ;
cwl:arguments ( "AnnotSV" "-bedtools" "/usr/bin/bedtools" "-outputDir" "$(runtime.outdir)" ) ;
cwl:arguments ( "AnnotSV" "-bedtools" "/usr/bin/bedtools" "-outputDir" "$(runtime.outdir)" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/annotsv-cwl:2.1"
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/annotsv-cwl:2.1"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 8000
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 8000
] .
cwl:in , ;
cwl:out ;
cwl:run .
rdfs:comment "output file name" ;
cwl:default "AnnotSV.tsv" ;
cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "-outputFile"
] ;
cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "-outputFile"
] ;
sld:type xsd:string , sld:null .
rdfs:comment "method used to merge allow multiallelic indels/snps" ;
cwl:default "none" ;
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "--merge"
] ;
sld:type [ sld:symbols , , , , , ;
sld:type sld:enum
] .
rdfs:comment "Minimum size of SVs to merge" ;
cwl:inputBinding [ CommandLineBinding:position
7 ] ;
sld:type xsd:int .
cwl:source .
cwl:in , , , ;
cwl:out ;
cwl:run .
rdfs:comment "snps vcf(s) for adding hom/het snp counts found within svs" ;
cwl:inputBinding [ CommandLineBinding:itemSeparator
"," ;
CommandLineBinding:position 5 ;
CommandLineBinding:prefix "-vcfFiles"
] ;
cwl:inputBinding [ CommandLineBinding:itemSeparator
"," ;
CommandLineBinding:position 5 ;
CommandLineBinding:prefix "-vcfFiles"
] ;
sld:type sld:null ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
rdfs:comment "Used to generate the output file name" ;
cwl:default "SURVIVOR-sv-merged.vcf" ;
cwl:inputBinding [ CommandLineBinding:position
8 ] ;
sld:type xsd:string , sld:null .
rdfs:comment "output vcf file name" ;
cwl:default "bcftools_merged.vcf.gz" ;
cwl:inputBinding [ CommandLineBinding:position 5 ;
CommandLineBinding:prefix "--output"
] ;
sld:type xsd:string , sld:null .
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "--filtering_frequency"
] ;
sld:type xsd:double , sld:null .