@prefix LoadListingRequirement: .
@prefix CommandLineBinding: .
@prefix xsd: .
@prefix DockerRequirement: .
@prefix Workflow: .
@prefix sld: .
@prefix rdfs: .
@prefix CommandOutputBinding: .
@prefix ns2: .
@prefix ns1: .
@prefix NetworkAccess: .
@prefix rdf: .
@prefix cwl: .
@prefix ResourceRequirement: .
cwl:in , , , , ;
cwl:out ;
cwl:run .
rdfs:comment "Filtered output of formatted and annotated Star Fusion and arriba results" ;
cwl:outputSource ;
sld:type cwl:File , sld:null .
rdfs:comment "gzipped rsem gene expression file" ;
sld:type cwl:File .
a cwl:File ;
cwl:path .
cwl:source .
rdfs:comment "Sample name to apply. Ought to be one from some kind of clinical database" ;
sld:type xsd:string .
a cwl:Workflow ;
ns1:original_cwlVersion "v1.0" ;
Workflow:steps , , ;
cwl:cwlVersion ;
cwl:hints [ a ns2:sevenbridges.comAWSInstanceType ;
rdfs:comment "Chosen for speed and lower cost"
] ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:MultipleInputFeatureRequirement ] .
cwl:source .
cwl:source .
cwl:outputBinding [ CommandOutputBinding:glob
"*.tsv" ] ;
sld:type cwl:File .
a cwl:CommandLineTool ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 1 ;
CommandLineBinding:shellQuote false ;
cwl:valueFrom "-I pigz -xf $(inputs.genome_tar.path) && /opt/FusionAnnotator/FusionAnnotator --genome_lib_dir ./$(inputs.genome_untar_path) --annotate $(inputs.input_fusion_file.path) --fusion_name_col $(inputs.col_num) > $(inputs.output_basename).annotated.tsv"
]
) ;
cwl:baseCommand ( "tar" ) ;
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:DockerRequirement ;
DockerRequirement:dockerPull "pgc-images.sbgenomics.com/d3b-bixu/fusionannotator:0.1.1"
] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 4 ;
ResourceRequirement:ramMin 8000
] ;
cwl:requirements [ a cwl:ShellCommandRequirement ] .
rdfs:comment "column number in file of fusion name, use 24 for arriba v1.1, 30 for v2" ;
cwl:default 30 ;
sld:type xsd:int , sld:null .
cwl:source .
sld:type cwl:File .
rdfs:comment "Fusion file formatted by format_fusion_file.cwl, likely from arriba input" ;
sld:type cwl:File .
sld:type xsd:string .
a cwl:CommandLineTool ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 1 ;
CommandLineBinding:shellQuote false ;
cwl:valueFrom "--outputfile $(inputs.sample_name).arriba_formatted.tsv"
]
) ;
cwl:baseCommand ( "Rscript" "formatArribaFusionCalls.R" ) ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , ;
cwl:outputs ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "pgc-images.sbgenomics.com/d3b-bixu/annofuse:0.92.0"
] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 2 ;
ResourceRequirement:ramMin 4000
] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "suppressPackageStartupMessages(library(\"readr\"))\nsuppressPackageStartupMessages(library(\"dplyr\"))\nsuppressPackageStartupMessages(library(\"optparse\"))\n\n\noption_list <- list(\n make_option(c(\"-f\", \"--fusionfile\"),type=\"character\",\n help=\"Fusion calls from Arriba\"),\n make_option(c(\"-t\", \"--tumorid\"), type=\"character\",\n help=\"KF tumor id\"),\n make_option(c(\"-o\",\"--outputfile\"),type=\"character\",\n help=\"Formatted fusion calls from [STARfusion | Arriba] (.TSV)\")\n)\n\n# Get command line options, if help option encountered print help and exit,\nopt <- parse_args(OptionParser(option_list=option_list))\ninputfile <- opt$fusionfile\ntumorid <- opt$tumorid\noutputfile <- opt$outputfile\n\ndf<-read_tsv(inputfile,col_types = readr::cols(breakpoint1 = readr::col_character(),breakpoint2 = readr::col_character()))\ndf$`gene1--gene2`<-paste(df$`#gene1`,df$gene2,sep=\"--\")\n\n\n# write to output file\nwrite.table(df,outputfile,sep=\"\\t\",quote=FALSE,row.names = FALSE)\n" ;
cwl:entryname "formatArribaFusionCalls.R"
]
] ;
cwl:requirements [ a cwl:ShellCommandRequirement ] .
cwl:source .
rdfs:comment "Arriba fusion file" ;
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "--fusionfile"
] ;
sld:type cwl:File .
cwl:outputBinding [ CommandOutputBinding:glob
"*.tsv" ] ;
sld:type cwl:File , sld:null .
rdfs:comment "0-based column number in file of fusion name, use 24 for arriba v1.1, 30 for v2" ;
cwl:default 30 ;
sld:type xsd:int , sld:null .
cwl:source .
rdfs:comment "Output from arriba, usually extension arriba.fusions.tsv" ;
sld:type cwl:File .
rdfs:comment "arriba fusion file formatted by format_fusion_file.cwl, and annotated by fusion_annotator.cwl" ;
sld:type cwl:File .
rdfs:comment "String to use as basename for outputs" ;
sld:type xsd:string .
cwl:source .
cwl:in , ;
cwl:out ;
cwl:run .
rdfs:comment "This is what the path will be when genome_tar is unpackaged" ;
cwl:default "GRCh38_v39_CTAT_lib_Mar242022.CUSTOM" ;
sld:type xsd:string , sld:null .
cwl:source .
rdfs:comment "Output from STAR Fusion, usually extension STAR.fusion_predictions.abridged.coding_effect.tsv" ;
sld:type cwl:File .
cwl:source .
cwl:source .
sld:type xsd:string .
cwl:outputBinding [ CommandOutputBinding:glob
"*.tsv" ] ;
sld:type cwl:File .
cwl:source .
a cwl:CommandLineTool ;
ns1:original_cwlVersion "v1.0" ;
cwl:arguments ( [ CommandLineBinding:position 1 ;
CommandLineBinding:shellQuote false ;
cwl:valueFrom "A_CT=`wc -l $(inputs.arriba_formatted_fusions.path) | cut -f 1 -d \" \"`\nS_CT=`wc -l $(inputs.starfusion_formatted_fusions.path) | cut -f 1 -d \" \"`\nif [ $A_CT -eq 1 ] && [ $S_CT -eq 1 ]; then\n echo \"Both inputs are empty, will skip processing as there no fusions.\" >&2;\n exit 0;\nfi\nRscript annoFusePerSample.R --fusionfileArriba $(inputs.arriba_formatted_fusions.path) --fusionfileStarFusion $(inputs.starfusion_formatted_fusions.path) --expressionFile $(inputs.rsem_expr_file.path) --tumorID $(inputs.sample_name) --outputfile $(inputs.output_basename).annoFuse_filter.tsv"
]
) ;
cwl:baseCommand () ;
cwl:cwlVersion ;
cwl:hints [ a ;
NetworkAccess:networkAccess true
] ;
cwl:hints [ a ;
LoadListingRequirement:loadListing
"deep_listing"
] ;
cwl:inputs , , , , ;
cwl:outputs ;
cwl:requirements [ a cwl:ResourceRequirement ;
ResourceRequirement:coresMin 4 ;
ResourceRequirement:ramMin 8000
] ;
cwl:requirements [ a cwl:InitialWorkDirRequirement ;
cwl:listing [ cwl:entry "library(\"annoFuse\")\n\nsuppressPackageStartupMessages(library(\"readr\"))\nsuppressPackageStartupMessages(library(\"tidyverse\"))\nsuppressPackageStartupMessages(library(\"reshape2\"))\nsuppressPackageStartupMessages(library(\"optparse\"))\n\noption_list <- list(\n make_option(c(\"-a\", \"--fusionfileArriba\"),type=\"character\",\n help=\"Formatted fusion calls from arriba\"),\n make_option(c(\"-s\", \"--fusionfileStarFusion\"),type=\"character\",\n help=\"Formatted fusion calls from starfusion\"),\n make_option(c(\"-e\", \"--expressionFile\"),type=\"character\",\n help=\"RSEM file for sample\"),\n make_option(c(\"-t\", \"--tumorID\"),type=\"character\",\n help=\"Sample name to rename column in RSEM FPKM column\"),\n make_option(c(\"-o\",\"--outputfile\"),type=\"character\",\n help=\"Formatted and filtered fusion calls from [STARfusion | Arriba] (.TSV)\")\n)\n\n#read in caller results\nopt <- parse_args(OptionParser(option_list=option_list))\nArribainputfile <- opt$fusionfileArriba\nSTARFusioninputfile <- opt$fusionfileStarFusion\nexpressionFile<-opt$expressionFile\ntumorID<-opt$tumorID\noutputfile <- opt$outputfile\n\nstandardFusioncalls <- annoFuse::annoFuse_single_sample(\n # Example files are provided in extdata, at-least 1 fusionfile is required along \n # with its rsem expression file\n fusionfileArriba = Arribainputfile,\n fusionfileStarFusion = STARFusioninputfile,\n expressionFile = expressionFile,\n tumorID = tumorID,\n # multiple read flag values for filtering using FusionAnnotator values\n artifactFilter = \"GTEx_Recurrent|DGD_PARALOGS|Normal|BodyMap\",\n # keep all in-frame , frameshift and other types of Fusion_Type\n readingFrameFilter = \"in-frame|frameshift|other\",\n # keep all fusions with atleast 1 junction read support\n junctionReadCountFilter = 1,\n # keep only fusions where spanningFragCount-junctionReadCountFilter less than equal to 100\n spanningFragCountFilter = 100,\n # filter read throughs\n readthroughFilter = FALSE\n)\n\n\n# write to outputfile\nwrite.table(standardFusioncalls,outputfile,sep=\"\\t\",quote=FALSE,row.names = FALSE)\n" ;
cwl:entryname "annoFusePerSample.R"
]
] ;
cwl:requirements [ a cwl:DockerRequirement ;
DockerRequirement:dockerPull "pgc-images.sbgenomics.com/d3b-bixu/annofuse:0.92.0"
] ;
cwl:requirements [ a cwl:ShellCommandRequirement ] ;
cwl:requirements [ a cwl:InlineJavascriptRequirement ] .
rdfs:comment "Tar ball with fusion_annot_lib.idx and blast_pairs.idx from STAR-Fusion CTAT Genome lib" ;
ns2:sevenbridges.comsuggestedValue
;
sld:type cwl:File .
cwl:in , , , , ;
cwl:out ;
cwl:run .
rdfs:comment "gzipped rsem gene expression file" ;
sld:type cwl:File .
sld:type xsd:string .
cwl:source .
rdfs:comment "STARFusion file formatted by format_fusion_file.cwl" ;
sld:type cwl:File .
rdfs:comment "Sample name" ;
cwl:inputBinding [ CommandLineBinding:position 1 ;
CommandLineBinding:prefix "--tumorid"
] ;
sld:type xsd:string .
rdfs:comment "This is what the path will be when genome_tar is unpackaged" ;
cwl:default "GRCh38_v39_CTAT_lib_Mar242022.CUSTOM" ;
sld:type xsd:string , sld:null .