WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl

   1 cwlVersion: v1.1
   2 class: Workflow
   3 label: Scattered variant calling workflow
   4
   5 requirements:
   6   SubworkflowFeatureRequirement: {}
   7   ScatterFeatureRequirement: {}
   8
   9 inputs:
  10   bam:
  11     type: File
  12     format: edam:format_2572 # BAM
  13     label: Indexed sorted BAM with labeled duplicates
  14     secondaryFiles:
  15       - .bai
  16   reference:
  17     type: File
  18     format: edam:format_1929 # FASTA
  19     label: Reference genome
  20     secondaryFiles:
  21       - .amb
  22       - .ann
  23       - .bwt
  24       - .pac
  25       - .sa
  26       - .fai
  27       - ^.dict
  28   fullintervallist:
  29     type: File
  30     label: Full list of intervals to operate over
  31   sample:
  32     type: string
  33     label: Sample Name
  34   knownsites:
  35     type: File
  36     format: edam:format_3016 # VCF
  37     label: VCF of known polymorphic sites for BQSR
  38     secondaryFiles:
  39       - .tbi
  40   scattercount:
  41     type: string
  42     label: Desired split for variant calling
  43
  44 outputs:
  45   gatheredgvcf:
  46     type: File
  47     format: edam:format_3016 # GVCF
  48     label: Gathered GVCF
  49     secondaryFiles:
  50       - .tbi
  51     outputSource: gather-GVCFs/gatheredgvcf
  52
  53 steps:
  54   splitintervals:
  55     run: gatk-splitintervals.cwl
  56     in:
  57       reference: reference
  58       fullintervallist: fullintervallist
  59       sample: sample
  60       scattercount: scattercount
  61     out: [intervalfiles]
  62
  63   recal-haplotypecaller:
  64     run: gatk-wf-with-interval.cwl
  65     scatter: intervallist
  66     in:
  67       bam: bam
  68       reference: reference
  69       sample: sample
  70       knownsites: knownsites
  71       intervallist: splitintervals/intervalfiles
  72     out: [gvcf]
  73
  74   gather-GVCFs:
  75     run: gather-array-vcf.cwl
  76     in:
  77       gvcfarray: recal-haplotypecaller/gvcf
  78       sample: sample
  79       reference: reference
  80     out: [gatheredgvcf]
  81
  82 s:codeRepository: https://github.com/arvados/arvados-tutorial
  83 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
  84
  85 $namespaces:
  86  s: https://schema.org/
  87  edam: http://edamontology.org/
  88
  89 #$schemas:
  90 # - https://schema.org/version/latest/schema.rdf
  91 # - http://edamontology.org/EDAM_1.18.owl