@prefix LoadListingRequirement: .
@prefix CommandLineBinding: .
@prefix xsd: .
@prefix DockerRequirement: .
@prefix Workflow: .
@prefix sld: .
@prefix rdfs: .
@prefix CommandOutputBinding: .
@prefix ns1: .
@prefix NetworkAccess: .
@prefix rdf: .
@prefix Dirent: .
@prefix cwl: .
a cwl:Workflow ;
rdfs:comment " Workflow to merge a large number of maf files into a single consensus maf file for use with GetBaseCountsMultiSample " ;
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:ScatterFeatureRequirement ] ;
cwl:requirements [ a cwl:SubworkflowFeatureRequirement ] ;
cwl:requirements [ a cwl:StepInputExpressionRequirement ] ;
cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] .
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
cwl:default "output.maf" ;
sld:type xsd:string .
a cwl:CommandLineTool ;
rdfs:comment " Special case of concat-tables.py to use with concatenating maf files for use with https://github.com/zengzheng123/GetBaseCountsMultiSample\nNote that these columns are required by GetBaseCountsMultiSample: ( as per https://github.com/zengzheng123/GetBaseCountsMultiSample/blob/master/GetBaseCountsMultiSample.cpp#L771-L801 )\nHugo_Symbol, Chromosome, Start_Position, End_Position, Reference_Allele, Tumor_Seq_Allele1, Tumor_Seq_Allele2, Tumor_Sample_Barcode, Matched_Norm_Sample_Barcode, t_ref_count, t_alt_count, n_ref_count, n_alt_count, Variant_Classification\nHowever, only these columns appear to influence the output:\nHugo_Symbol Chromosome Start_Position End_Position Variant_Classification Reference_Allele Tumor_Seq_Allele1 Tumor_Seq_Allele2\nusually in a column order matching this; grep -v '#' Proj_08390_G.muts.maf | cut -f1,5,6,7,9,11,12,13,16,17,41,42,44,45 > Proj_08390_G.cols.muts.maf\nalso, if present these columns also appear in output but seem to be optional\nCenter\tNCBI_Build\nsee also; https://docs.gdc.cancer.gov/Data/File_Formats/MAF_Format/ " ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "-o" ;
cwl:valueFrom "$(inputs.output_filename)"
]
[ CommandLineBinding:position 2 ;
cwl:valueFrom "${ return \"inputs_dir\" }"
]
) ;
cwl:baseCommand ( "concat-tables.py" "--dir" "--na-str" "." "--keep-cols" "Hugo_Symbol" "Chromosome" "Start_Position" "End_Position" "Variant_Classification" "Reference_Allele" "Tumor_Seq_Allele1" "Tumor_Seq_Allele2" "Center" "NCBI_Build" "--na-cols" "Tumor_Sample_Barcode" "Matched_Norm_Sample_Barcode" "t_ref_count" "t_alt_count" "n_ref_count" "n_alt_count" "--comments" "--no-carriage-returns" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs , ;
cwl:outputs ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ Dirent:writable true ;
cwl:entry "${\n for (var i = 0; i < inputs.input_files.length; i++) {\n inputs.input_files[i].basename = inputs.input_files[i].basename + \".\" + i;\n }\n return {class: 'Directory', listing: inputs.input_files};\n}" ;
cwl:entryname "inputs_dir"
]
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mskcc/helix_filters_01:21.4.1"
] .
cwl:source .
cwl:inputBinding [ CommandLineBinding:position
1 ] ;
sld:type cwl:File .
sld:type [ sld:items cwl:File ;
sld:type sld:array
] .
cwl:in ;
cwl:out ;
cwl:run .
cwl:in ;
cwl:out ;
cwl:run .
cwl:outputBinding [ CommandOutputBinding:glob
"$(inputs.output_filename)" ] ;
sld:type cwl:File .
cwl:outputBinding [ CommandOutputBinding:glob
"dedup.maf" ] ;
sld:type cwl:File .
a cwl:CommandLineTool ;
rdfs:comment "Remove duplicate lines from the maf file\nExpected columns are in this order;\n1 Hugo_Symbol 2 Center 3 NCBI_Build 4 Chromosome 5 Start_Position 6 End_Position 7 Variant_Classification 8 Reference_Allele 9 Tumor_Seq_Allele1 10 Tumor_Seq_Allele2 11 n_alt_count 12 Matched_Norm_Sample_Barcode 13 t_alt_count 14 t_ref_count 15 n_ref_count 16 Tumor_Sample_Barcode\nNOTE: actually this order might be wrong but all that matters is that Chromosome and Start_Position are columns 4 and 5 " ;
ns1:original_cwlVersion "v1.0" ;
cwl:baseCommand ( "bash" "run.sh" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:inputs ;
cwl:outputs ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "set -euo pipefail\ngrep '#' \"$1\" > comments || touch comments\ngrep '^Hugo_Symbol' \"$1\" > header || touch header\ngrep -v '^Hugo_Symbol' \"$1\" | grep -v '#' | sort -u > body.dedup\nsort -V -k4,4 -k5,5n body.dedup > body.sorted\ncat comments > dedup.maf\ncat header >> dedup.maf\ncat body.sorted >> dedup.maf" ;
cwl:entryname "run.sh"
]
] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "mskcc/helix_filters_01:21.4.1"
] .
cwl:source .
cwl:outputSource ;
sld:type cwl:File .