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   knownsites:
  26     type: File
  27     format: edam:format_3016 # VCF
  28     label: VCF of known polymorphic sites for BQSR
  29     secondaryFiles:
  30       - .tbi
  31   scattercount:
  32     type: string
  33     label: Desired split for variant calling
  34   clinvarvcf:
  35     type: File
  36     format: edam:format_3016 # VCF
  37     label: Reference VCF for ClinVar
  38   reportfunc:
  39     type: File
  40     label: Function used to create HTML report
  41   headhtml:
  42     type: File
  43     format: edam:format_1964 # HTML
  44     label: Header for HTML report
  45   tailhtml:
  46     type: File
  47     format: edam:format_1964 # HTML
  48     label: Footer for HTML report
  49
  50 outputs:
  51   gvcf:
  52     type: File[]
  53     outputSource: bwamem-gatk-report/gvcf
  54     format: edam:format_3016 # GVCF
  55     label: GVCFs generated from GATK
  56
  57   report:
  58     type: File[]
  59     outputSource: bwamem-gatk-report/report
  60     format: edam:format_1964 # HTML
  61     label: ClinVar variant reports
  62
  63 steps:
  64   getfastq:
  65     run: ./helper/getfastq.cwl
  66     in:
  67       fastqdir: fastqdir
  68     out: [fastq1, fastq2, sample]
  69
  70   bwamem-gatk-report:
  71     run: ./helper/bwamem-gatk-report-wf.cwl
  72     scatter: [fastq1, fastq2, sample]
  73     scatterMethod: dotproduct
  74     in:
  75       fastq1: getfastq/fastq1
  76       fastq2: getfastq/fastq2
  77       reference: reference
  78       sample: getfastq/sample
  79       knownsites: knownsites
  80       scattercount: scattercount
  81       clinvarvcf: clinvarvcf
  82       reportfunc: reportfunc
  83       headhtml: headhtml
  84       tailhtml: tailhtml
  85     out: [qc-html,qc-zip,gvcf,report]
  86
  87 s:codeRepository: https://github.com/arvados/arvados-tutorial
  88 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
  89
  90 $namespaces:
  91  s: https://schema.org/
  92  edam: http://edamontology.org/
  93
  94 $schemas:
  95  - https://schema.org/version/latest/schema.rdf
  96  - http://edamontology.org/EDAM_1.18.owl