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   sample:
  26     type: string
  27     label: Sample name
  28   knownsites:
  29     type: File
  30     format: edam:format_3016 # VCF
  31     label: VCF of known polymorphic sites for BQSR
  32     secondaryFiles:
  33       - .tbi
  34   scattercount:
  35     type: string
  36     label: Desired split for variant calling
  37   clinvarvcf:
  38     type: File
  39     format: edam:format_3016 # VCF
  40     label: Reference VCF for ClinVar
  41   reportfunc:
  42     type: File
  43     label: Function used to create HTML report
  44   headhtml:
  45     type: File
  46     format: edam:format_1964 # HTML
  47     label: Header for HTML report
  48   tailhtml:
  49     type: File
  50     format: edam:format_1964 # HTML
  51     label: Footer for HTML report
  52
  53 outputs:
  54   gvcf:
  55     type: File[]
  56     outputSource: bwamem-gatk-report/gvcf
  57     format: edam:format_3016 # GVCF
  58     label: GVCFs generated from sets of fastqs
  59
  60   report:
  61     type: File[]
  62     outputSource: bwamem-gatk-report/report
  63     format: edam:format_1964 # HTML
  64     label: ClinVar variant reports
  65
  66 steps:
  67   getfastq:
  68     run: ./helper/getfastq.cwl
  69     in:
  70       fastqdir: fastqdir
  71     out: [fastq1, fastq2, sample]
  72
  73   bwamem-gatk-report:
  74     run: ./helper/bwamem-gatk-report-wf.cwl
  75     scatter: [fastq1, fastq2, sample]
  76     scatterMethod: dotproduct
  77     in:
  78       fastq1: getfastq/fastq1
  79       fastq2: getfastq/fastq2
  80       reference: reference
  81       sample: getfastq/sample
  82       knownsites: knownsites
  83       scattercount: scattercount
  84       clinvarvcf: clinvarvcf
  85       reportfunc: reportfunc
  86       headhtml: headhtml
  87       tailhtml: tailhtml
  88     out: [qc-html,qc-zip,gvcf,report]
  89
  90 s:codeRepository: https://github.com/arvados/arvados-tutorial
  91 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
  92
  93 $namespaces:
  94  s: https://schema.org/
  95  edam: http://edamontology.org/
  96
  97 $schemas:
  98  - https://schema.org/version/latest/schema.rdf
  99  - http://edamontology.org/EDAM_1.18.owl