WGS-processing/cwl/wgs-processing-wf.cwl

   1 cwlVersion: v1.1
   2 class: Workflow
   3 label: WGS processing workflow scattered over samples
   4
   5 requirements:
   6   - class: SubworkflowFeatureRequirement
   7   - class: ScatterFeatureRequirement
   8
   9 inputs:
  10   fastqdir:
  11     type: Directory
  12     label: Directory of paired FASTQ files
  13   reference:
  14     type: File
  15     format: edam:format_1929 # FASTA
  16     label: Reference genome
  17     secondaryFiles:
  18       - .amb
  19       - .ann
  20       - .bwt
  21       - .pac
  22       - .sa
  23       - .fai
  24       - ^.dict
  25   fullintervallist:
  26     type: File
  27   knownsites:
  28     type: File
  29     format: edam:format_3016 # VCF
  30     label: VCF of known polymorphic sites for BQSR
  31     secondaryFiles:
  32       - .tbi
  33   scattercount:
  34     type: string
  35     label: Desired split for variant calling
  36   clinvarvcf:
  37     type: File
  38     format: edam:format_3016 # VCF
  39     label: Reference VCF for ClinVar
  40   reportfunc:
  41     type: File
  42     label: Function used to create HTML report
  43   headhtml:
  44     type: File
  45     format: edam:format_2331 # HTML
  46     label: Header for HTML report
  47   tailhtml:
  48     type: File
  49     format: edam:format_2331 # HTML
  50     label: Footer for HTML report
  51
  52 outputs:
  53   gvcf:
  54     type: File[]
  55     outputSource: bwamem-gatk-report/gvcf
  56     format: edam:format_3016 # GVCF
  57     label: GVCFs generated from GATK
  58   report:
  59     type: File[]
  60     outputSource: bwamem-gatk-report/report
  61     format: edam:format_2331 # HTML
  62     label: ClinVar variant reports
  63
  64 steps:
  65   getfastq:
  66     run: ./helper/getfastq.cwl
  67     in:
  68       fastqdir: fastqdir
  69     out: [fastq1, fastq2, sample]
  70
  71   bwamem-gatk-report:
  72     run: ./helper/bwamem-gatk-report-wf.cwl
  73     scatter: [fastq1, fastq2, sample]
  74     scatterMethod: dotproduct
  75     in:
  76       fastq1: getfastq/fastq1
  77       fastq2: getfastq/fastq2
  78       reference: reference
  79       fullintervallist: fullintervallist
  80       sample: getfastq/sample
  81       knownsites: knownsites
  82       scattercount: scattercount
  83       clinvarvcf: clinvarvcf
  84       reportfunc: reportfunc
  85       headhtml: headhtml
  86       tailhtml: tailhtml
  87     out: [qc-html,qc-zip,gvcf,report]
  88
  89 s:codeRepository: https://github.com/arvados/arvados-tutorial
  90 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
  91
  92 $namespaces:
  93  s: https://schema.org/
  94  edam: http://edamontology.org/
  95
  96 #$schemas:
  97 # - https://schema.org/version/latest/schema.rdf
  98 # - http://edamontology.org/EDAM_1.18.owl