@prefix LoadListingRequirement: .
@prefix SchemaDefRequirement: .
@prefix CommandLineBinding: .
@prefix xsd: .
@prefix DockerRequirement: .
@prefix Workflow: .
@prefix sld: .
@prefix rdfs: .
@prefix CommandOutputBinding: .
@prefix ns1: .
@prefix NetworkAccess: .
@prefix rdf: .
@prefix ScatterMethod: .
@prefix cwl: .
@prefix ResourceRequirement: .
cwl:outputSource ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source ;
cwl:valueFrom "$(self.readgroup)" .
rdfs:comment "bwa-indexed reference file" ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:source ;
cwl:valueFrom "$(self.sequence.hasOwnProperty('bam')? self.sequence.bam : null)" .
sld:type cwl:File .
rdfs:comment "Minimum overlap for removal without pair overlap." ;
sld:type xsd:int .
sld:type , sld:null .
cwl:inputBinding [ CommandLineBinding:prefix
"-g" ] ;
sld:type xsd:string .
rdfs:label "trimming adapter file and parameters" ;
sld:fields , ;
sld:type sld:record .
cwl:default "MarkedSorted.bam" ;
cwl:inputBinding [ CommandLineBinding:position
3 ] ;
sld:type xsd:string , sld:null .
cwl:source ;
cwl:valueFrom "$(self.sequence.hasOwnProperty('fastq2')? self.sequence.fastq2 : null)" .
rdfs:comment "One or more databases of known polymorphic sites used to exclude regions around known polymorphisms from analysis." ;
cwl:secondaryFiles [] ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
sld:type , .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.bam.nameroot).mark_dups_metrics.txt" ] ;
sld:type cwl:File .
cwl:in , , ;
cwl:out ;
cwl:run ;
cwl:scatter ( ) ;
cwl:scatterMethod ScatterMethod:dotproduct .
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "-I"
] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:prefix
"-2" ] ;
sld:type cwl:File , sld:null .
cwl:secondaryFiles [] ;
sld:type [ cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "--known-sites"
] ;
sld:items cwl:File ;
sld:type sld:array
] .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_name).bam" ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
a cwl:CommandLineTool ;
rdfs:comment "Due to workflow runner limitations, use sequence_align_and_tag_adapter.cwl subworkflow to call this" ;
rdfs:label "align with bwa_mem and tag" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 5 ;
CommandLineBinding:prefix "-n" ;
cwl:valueFrom "$(runtime.cores)"
]
) ;
cwl:baseCommand ( "/bin/bash" "sequence_alignment_helper.sh" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:SchemaDefRequirement ;
SchemaDefRequirement:types ,
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/alignment_helper-cwl:1.1.0"
] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "set -o pipefail\nset -o errexit\nset -o nounset\n\nRUN_TRIMMING=\"false\"\n\nwhile getopts \"b:?1:?2:?g:r:n:t:?o:?\" opt; do\n case \"$opt\" in\n b)\n MODE=bam\n BAM=\"$OPTARG\"\n ;;\n 1)\n MODE=fastq\n FASTQ1=\"$OPTARG\"\n ;;\n 2) \n MODE=fastq\n FASTQ2=\"$OPTARG\"\n ;;\n g)\n READGROUP=\"$OPTARG\"\n ;;\n r)\n REFERENCE=\"$OPTARG\"\n ;;\n n)\n NTHREADS=\"$OPTARG\"\n ;;\n t)\n RUN_TRIMMING=\"true\"\n TRIMMING_ADAPTERS=\"$OPTARG\"\n ;;\n o)\n RUN_TRIMMING=\"true\"\n TRIMMING_ADAPTER_MIN_OVERLAP=\"$OPTARG\"\n ;;\n esac\ndone\n\nif [[ \"$MODE\" == 'fastq' ]]; then\n if [[ \"$RUN_TRIMMING\" == 'false' ]]; then\n /usr/local/bin/bwa mem -K 100000000 -t \"$NTHREADS\" -Y -R \"$READGROUP\" \"$REFERENCE\" \"$FASTQ1\" \"$FASTQ2\" | /usr/local/bin/samblaster -a --addMateTags | /opt/samtools/bin/samtools view -b -S /dev/stdin\n else\n /opt/flexbar/flexbar --adapters \"$TRIMMING_ADAPTERS\" --reads \"$FASTQ1\" --reads2 \"$FASTQ2\" --adapter-trim-end LTAIL --adapter-min-overlap \"$TRIMMING_ADAPTER_MIN_OVERLAP\" --adapter-error-rate 0.1 --max-uncalled 300 --stdout-reads \\\n | /usr/local/bin/bwa mem -K 100000000 -t \"$NTHREADS\" -Y -p -R \"$READGROUP\" \"$REFERENCE\" /dev/stdin | /usr/local/bin/samblaster -a --addMateTags | /opt/samtools/bin/samtools view -b -S /dev/stdin\n fi\nfi\nif [[ \"$MODE\" == 'bam' ]]; then\n if [[ \"$RUN_TRIMMING\" == 'false' ]]; then\n /usr/bin/java -Xmx4g -jar /opt/picard/picard.jar SamToFastq I=\"$BAM\" INTERLEAVE=true INCLUDE_NON_PF_READS=true FASTQ=/dev/stdout | /usr/local/bin/bwa mem -K 100000000 -t \"$NTHREADS\" -Y -p -R \"$READGROUP\" \"$REFERENCE\" /dev/stdin | /usr/local/bin/samblaster -a --addMateTags | /opt/samtools/bin/samtools view -b -S /dev/stdin\n else\n /usr/bin/java -Xmx4g -jar /opt/picard/picard.jar SamToFastq I=\"$BAM\" INTERLEAVE=true INCLUDE_NON_PF_READS=true FASTQ=/dev/stdout \\\n | /opt/flexbar/flexbar --adapters \"$TRIMMING_ADAPTERS\" --reads - --interleaved --adapter-trim-end LTAIL --adapter-min-overlap \"$TRIMMING_ADAPTER_MIN_OVERLAP\" --adapter-error-rate 0.1 --max-uncalled 300 --stdout-reads \\\n | /usr/local/bin/bwa mem -K 100000000 -t \"$NTHREADS\" -Y -p -R \"$READGROUP\" \"$REFERENCE\" /dev/stdin | /usr/local/bin/samblaster -a --addMateTags | /opt/samtools/bin/samtools view -b -S /dev/stdin\n fi\nfi\n" ;
cwl:entryname "sequence_alignment_helper.sh"
]
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 8 ;
ResourceRequirement:ramMin 20000
] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] ;
cwl:stdout "refAlign.bam" .
a cwl:CommandLineTool ;
rdfs:label "sort BAM by name" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-t"
[ cwl:valueFrom "$(runtime.cores)" ]
"-m" "22G" "-n" "-o"
[ cwl:valueFrom "$(inputs.bam.nameroot).NameSorted.bam" ]
) ;
cwl:baseCommand ( "/usr/bin/sambamba" "sort" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 8 ;
ResourceRequirement:ramMin 26000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mgibio/sambamba-cwl:0.6.4"
] .
cwl:default "final" ;
sld:type xsd:string , sld:null .
rdfs:label "sequence data with readgroup information" ;
sld:fields , ;
sld:type sld:record .
cwl:in , , , ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.bam.basename)" ] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type cwl:File .
cwl:in , ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.bam.nameroot).NameSorted.bam" ] ;
sld:type cwl:File .
rdfs:comment "bwa-indexed reference file" ;
cwl:inputBinding [ CommandLineBinding:position 4 ;
CommandLineBinding:prefix "-r"
] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:source .
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "-R"
] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
a cwl:CommandLineTool ;
rdfs:label "apply BQSR" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-O"
[ cwl:valueFrom "$(runtime.outdir)/$(inputs.output_name).bam" ]
"--static-quantized-quals" "10" "--static-quantized-quals" "20" "--static-quantized-quals" "30"
) ;
cwl:baseCommand ( "/gatk/gatk" "--java-options" "-Xmx16g" "ApplyBQSR" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:ramMin 18000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "broadinstitute/gatk:4.1.8.1"
] .
cwl:source .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type cwl:File .
sld:type cwl:File .
sld:type sld:null ;
sld:type [ sld:items xsd:string ;
sld:type sld:array
] .
cwl:source .
cwl:source .
cwl:inputBinding [ cwl:valueFrom "$( ['-t', self.adapters.path, '-o', self.min_overlap] )" ] ;
sld:type , sld:null .
cwl:inputBinding [ CommandLineBinding:prefix
"-b" ] ;
sld:type cwl:File , sld:null .
cwl:source .
cwl:source .
sld:fields , ;
sld:type sld:record .
cwl:in ;
cwl:out ;
cwl:run .
cwl:default "queryname" ;
cwl:inputBinding [ CommandLineBinding:position
5 ] ;
sld:type xsd:string .
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "-I"
] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
cwl:source .
cwl:default "chr11" , "chr10" , "chr19" , "chr18" , "chr5" , "chr9" , "chr17" , "chr14" , "chr21" , "chr6" , "chr2" , "chr13" , "chr4" , "chr1" , "chr8" , "chr16" , "chr3" , "chr20" , "chr7" , "chr15" , "chr22" , "chr12" ;
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type [ cwl:inputBinding [ CommandLineBinding:prefix
"-L" ] ;
sld:items xsd:string ;
sld:type sld:array
] .
sld:type xsd:string .
cwl:source .
a cwl:CommandLineTool ;
rdfs:label "Mark duplicates and Sort" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 2 ;
cwl:valueFrom "$(runtime.cores)"
]
[ CommandLineBinding:position 4 ;
cwl:valueFrom "$(inputs.bam.nameroot).mark_dups_metrics.txt"
]
) ;
cwl:baseCommand ( "/bin/bash" "markduplicates_helper.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 "mgibio/mark_duplicates-cwl:1.0.1"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 8 ;
ResourceRequirement:ramMin 40000
] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "set -o pipefail\nset -o errexit\n\ndeclare MD_BARCODE_TAG\nif [ ! -z \"$6\" ]; then\n MD_BARCODE_TAG=\"BARCODE_TAG=$6\"\n/usr/bin/java -Xmx16g -jar /opt/picard/picard.jar MarkDuplicates I=$1 O=/dev/stdout ASSUME_SORT_ORDER=$5 METRICS_FILE=$4 QUIET=true COMPRESSION_LEVEL=0 VALIDATION_STRINGENCY=LENIENT \"$MD_BARCODE_TAG\" | /usr/bin/sambamba sort -t $2 -m 18G -o $3 /dev/stdin\nelse\n /usr/bin/java -Xmx16g -jar /opt/picard/picard.jar MarkDuplicates I=$1 O=/dev/stdout ASSUME_SORT_ORDER=$5 METRICS_FILE=$4 QUIET=true COMPRESSION_LEVEL=0 VALIDATION_STRINGENCY=LENIENT | /usr/bin/sambamba sort -t $2 -m 18G -o $3 /dev/stdin\nfi\n" ;
cwl:entryname "markduplicates_helper.sh"
]
] .
rdfs:comment "Fasta file with adapters for removal that may contain N." ;
sld:type cwl:File .
sld:type , sld:null .
cwl:outputSource ;
sld:type cwl:File .
cwl:source .
sld:type cwl:File .
a cwl:CommandLineTool ;
rdfs:label "create BQSR table" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "-O"
[ cwl:valueFrom "$(runtime.outdir)/bqsr.table" ]
) ;
cwl:baseCommand ( "/gatk/gatk" "--java-options" "-Xmx16g" "BaseRecalibrator" ) ;
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 18000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "broadinstitute/gatk:4.1.8.1"
] .
cwl:source .
cwl:default "final" ;
sld:type xsd:string .
cwl:in , , , , , ;
cwl:out ;
cwl:run .
cwl:in , , , ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"refAlign.bam" ] ;
sld:type cwl:File .
a cwl:Workflow ;
rdfs:label "Raw sequence data to BQSR" ;
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:ScatterFeatureRequirement ] ;
cwl:requirements [ a cwl:SchemaDefRequirement ;
SchemaDefRequirement:types ,
] ;
cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] ;
cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] .
cwl:source ;
cwl:valueFrom "$(self.sequence.hasOwnProperty('fastq1')? self.sequence.fastq1 : null)" .
cwl:source .
sld:type [ sld:items ;
sld:type sld:array
] .
sld:type cwl:File .
sld:fields ;
sld:type sld:record .
cwl:outputBinding [ CommandOutputBinding:glob
"bqsr.table" ] ;
sld:type cwl:File .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position 2 ;
CommandLineBinding:prefix "-R"
] ;
cwl:secondaryFiles [] ;
cwl:secondaryFiles [] ;
sld:type cwl:File , xsd:string .
cwl:in ;
cwl:out ;
cwl:run .
cwl:inputBinding [ CommandLineBinding:position 3 ;
CommandLineBinding:prefix "-bqsr"
] ;
sld:type cwl:File .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.name).merged.bam" ] ;
sld:type cwl:File .
sld:type xsd:string .
cwl:inputBinding [ CommandLineBinding:prefix
"-1" ] ;
sld:type cwl:File , sld:null .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
a cwl:CommandLineTool ;
rdfs:label "samtools index" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "/usr/local/bin/samtools" "index" "$(runtime.outdir)/$(inputs.bam.basename)" "$(runtime.outdir)/$(inputs.bam.basename).bai"
[ CommandLineBinding:shellQuote false ;
cwl:valueFrom " && "
]
"cp" "$(inputs.bam.basename).bai" "$(runtime.outdir)/$(inputs.bam.nameroot).bai"
) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing "${ var f = inputs.bam; delete f.secondaryFiles; return f }"
] ;
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:ShellCommandRequirement ] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] .
a cwl:CommandLineTool ;
rdfs:label "Samtools: merge" ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( "$(inputs.name).merged.bam"
[ CommandLineBinding:prefix "--threads" ;
cwl:valueFrom "$(runtime.cores)"
]
) ;
cwl:baseCommand ( "/usr/local/bin/samtools" "merge" ) ;
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:coresMin 4 ;
ResourceRequirement:ramMin 8000
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "quay.io/biocontainers/samtools:1.11--h6270b1f_0"
] .
cwl:source .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_name)" ] ;
cwl:secondaryFiles [] ;
sld:type cwl:File .
rdfs:comment "the unaligned sequence data with readgroup information" ;
sld:type .
cwl:in ;
cwl:out , ;
cwl:run .
cwl:outputSource ;
sld:type cwl:File .
cwl:source .
a cwl:Workflow ;
rdfs:comment "Some workflow engines won't stage files in our nested structure, so parse it out here" ;
rdfs:label "adapter for sequence_align_and_tag" ;
ns1:original_cwlVersion "v1.0" ;
Workflow:steps ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs ,