@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: .
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:default "NOPREFIX" ;
cwl:inputBinding [ CommandLineBinding:position
-4 ] ;
sld:type xsd:string , sld:null .
cwl:inputBinding [ CommandLineBinding:position
-7 ] ;
sld:type xsd:string .
cwl:outputSource ;
sld:type cwl:File .
cwl:outputSource ;
sld:type cwl:File .
cwl:default 20 ;
cwl:inputBinding [ CommandLineBinding:position
-2 ] ;
sld:type xsd:int , sld:null .
cwl:source .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
-8 ] ;
sld:type cwl:File .
cwl:source .
cwl:in , , , , , ;
cwl:out , ;
cwl:run .
sld:type xsd:string .
a cwl:Workflow ;
rdfs:label "bam_readcount workflow" ;
ns1:original_cwlVersion "v1.0" ;
Workflow:steps , , ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , , , , , ;
cwl:outputs , , ;
cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] .
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "--variant"
] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
sld:type xsd:int , sld:null .
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:outputSource ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
-6 ] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:inputBinding [ CommandLineBinding:position
-5 ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "-R"
] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:outputBinding [ CommandOutputBinding:glob
"normalized.vcf.gz" ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:default 0 ;
cwl:inputBinding [ CommandLineBinding:position
-1 ] ;
sld:type xsd:int , sld:null .
cwl:source .
cwl:in , ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"decomposed.vcf.gz" ] ;
sld:type cwl:File .
a cwl:CommandLineTool ;
rdfs:label "run bam-readcount" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position -3 ;
cwl:valueFrom "$(runtime.outdir)"
]
) ;
cwl:baseCommand ( "/usr/bin/python" "bam_readcount_helper.py" ) ;
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 16000
] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] ;
cwl:requirements [ a cwl:ShellCommandRequirement ] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "#!/usr/bin/env python\n\nimport sys\nimport os\nfrom cyvcf2 import VCF\nimport tempfile\nimport csv\nfrom subprocess import Popen, PIPE\n\ndef generate_region_list(hash):\n fh = tempfile.NamedTemporaryFile('w', delete=False)\n writer = csv.writer(fh, delimiter='\\t')\n for chr, positions in hash.items():\n for pos in sorted(positions.keys()):\n writer.writerow([chr, pos, pos])\n fh.close()\n return fh.name\n\ndef filter_sites_in_hash(region_list, bam_file, ref_fasta, prefixed_sample, output_dir, insertion_centric, map_qual, base_qual):\n bam_readcount_cmd = ['/usr/bin/bam-readcount', '-f', ref_fasta, '-l', region_list, '-w', '0', '-b', str(base_qual), '-q', str(map_qual)]\n if insertion_centric:\n bam_readcount_cmd.append('-i')\n output_file = os.path.join(output_dir, prefixed_sample + '_bam_readcount_indel.tsv')\n else:\n output_file = os.path.join(output_dir, prefixed_sample + '_bam_readcount_snv.tsv')\n bam_readcount_cmd.append(bam_file)\n execution = Popen(bam_readcount_cmd, stdout=PIPE, stderr=PIPE)\n stdout, stderr = execution.communicate()\n if execution.returncode == 0:\n with open(output_file, 'wb') as output_fh:\n output_fh.write(stdout)\n else:\n sys.exit(stderr)\n\n#initializing these with default values\nmin_base_qual = 20\nmin_mapping_qual = 0\n\nif len(sys.argv) == 7:\n (script_name, vcf_filename, sample, ref_fasta, bam_file, prefix, output_dir)= sys.argv\nelif len(sys.argv) == 8:\n (script_name, vcf_filename, sample, ref_fasta, bam_file, prefix, output_dir, min_base_qual)= sys.argv\nelif len(sys.argv) == 9: #elif instead of else for explicit safety\n (script_name, vcf_filename, sample, ref_fasta, bam_file, prefix, output_dir, min_base_qual, min_mapping_qual)= sys.argv\n\nif prefix == 'NOPREFIX':\n prefixed_sample = sample\nelse:\n prefixed_sample = '_'.join([prefix, sample])\n\nvcf_file = VCF(vcf_filename)\nsample_index = vcf_file.samples.index(sample)\n\nrc_for_indel = {}\nrc_for_snp = {}\nfor variant in vcf_file:\n ref = variant.REF\n chr = variant.CHROM\n start = variant.start\n end = variant.end\n pos = variant.POS\n for var in variant.ALT:\n if len(ref) > 1 or len(var) > 1:\n #it's an indel or mnp\n if len(ref) == len(var) or (len(ref) > 1 and len(var) > 1):\n sys.stderr.write(\"Complex variant or MNP will be skipped: %s\\t%s\\t%s\\t%s\\n\" % (chr, pos, ref , var))\n continue\n elif len(ref) > len(var):\n #it's a deletion\n pos += 1\n unmodified_ref = ref\n ref = unmodified_ref[1]\n var = \"-%s\" % unmodified_ref[1:]\n else:\n #it's an insertion\n var = \"+%s\" % var[1:]\n if chr not in rc_for_indel:\n rc_for_indel[chr] = {}\n if pos not in rc_for_indel[chr]:\n rc_for_indel[chr][pos] = {}\n if ref not in rc_for_indel[chr][pos]:\n rc_for_indel[chr][pos][ref] = {}\n rc_for_indel[chr][pos][ref] = variant\n else:\n #it's a SNP\n if chr not in rc_for_snp:\n rc_for_snp[chr] = {}\n if pos not in rc_for_snp[chr]:\n rc_for_snp[chr][pos] = {}\n if ref not in rc_for_snp[chr][pos]:\n rc_for_snp[chr][pos][ref] = {}\n rc_for_snp[chr][pos][ref] = variant\n\nif len(rc_for_snp.keys()) > 0:\n region_file = generate_region_list(rc_for_snp)\n filter_sites_in_hash(region_file, bam_file, ref_fasta, prefixed_sample, output_dir, False, min_mapping_qual, min_base_qual)\nelse:\n output_file = os.path.join(output_dir, prefixed_sample + '_bam_readcount_snv.tsv')\n open(output_file, 'w').close()\n\nif len(rc_for_indel.keys()) > 0:\n region_file = generate_region_list(rc_for_indel)\n filter_sites_in_hash(region_file, bam_file, ref_fasta, prefixed_sample, output_dir, True, min_mapping_qual, min_base_qual)\nelse:\n output_file = os.path.join(output_dir, prefixed_sample + '_bam_readcount_indel.tsv')\n open(output_file, 'w').close()\n" ;
cwl:entryname "bam_readcount_helper.py"
]
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/bam_readcount_helper-cwl:1.1.1"
] ;
cwl:stdout "$(inputs.sample)_bam_readcount.tsv" .
a cwl:CommandLineTool ;
rdfs:label "run vt decompose" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-s" "-o"
[ cwl:valueFrom "$(runtime.outdir)/decomposed.vcf.gz" ]
) ;
cwl:baseCommand ( "vt" "decompose" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/vt:0.57721--hf74b74d_1"
] .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"${\n var name = \"_bam_readcount_snv.tsv\";\n if (inputs.prefix.equals(\"NOPREFIX\")) {\n name = inputs.sample + name;\n }\n else {\n name = inputs.prefix + \"_\" + inputs.sample + name;\n }\n return name;\n}\n" ] ;
sld:type cwl:File .
cwl:source .
a cwl:CommandLineTool ;
rdfs:label "Normalize variants" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-o"
[ cwl:valueFrom "$(runtime.outdir)/normalized.vcf.gz" ]
) ;
cwl:baseCommand ( "/usr/bin/java" "-Xmx8g" "-jar" "/opt/GenomeAnalysisTK.jar" "-T" "LeftAlignAndTrimVariants" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 9000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/gatk-cwl:3.6.0"
] .
sld:type xsd:int , sld:null .
cwl:outputBinding [ CommandOutputBinding:glob
"${\n var name = \"_bam_readcount_indel.tsv\";\n if (inputs.prefix.equals(\"NOPREFIX\")) {\n name = inputs.sample + name;\n }\n else {\n name = inputs.prefix + \"_\" + inputs.sample + name;\n }\n return name;\n}\n" ] ;
sld:type cwl:File .
cwl:in ;
cwl:out ;
cwl:run .