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   knownsites1:
  28     type: File
  29     format: edam:format_3016 # VCF
  30     label: VCF of known SNPS sites for BQSR
  31     secondaryFiles:
  32       - .idx
  33   knownsites2:
  34     type: File
  35     format: edam:format_3016 # VCF
  36     label: VCF of known indel sites for BQSR
  37     secondaryFiles:
  38       - .tbi
  39   scattercount:
  40     type: string
  41     label: Desired split for variant calling
  42   clinvarvcf:
  43     type: File
  44     format: edam:format_3016 # VCF
  45     label: Reference VCF for ClinVar
  46   reportfunc:
  47     type: File
  48     label: Function used to create HTML report
  49   headhtml:
  50     type: File
  51     format: edam:format_2331 # HTML
  52     label: Header for HTML report
  53   tailhtml:
  54     type: File
  55     format: edam:format_2331 # HTML
  56     label: Footer for HTML report
  57
  58 outputs:
  59   gvcf:
  60     type: File[]
  61     outputSource: bwamem-gatk-report/gvcf
  62     format: edam:format_3016 # GVCF
  63     label: GVCFs generated from GATK
  64   report:
  65     type: File[]
  66     outputSource: bwamem-gatk-report/report
  67     format: edam:format_2331 # HTML
  68     label: ClinVar variant reports
  69
  70 steps:
  71   getfastq:
  72     run: ./helper/getfastq.cwl
  73     in:
  74       fastqdir: fastqdir
  75     out: [fastq1, fastq2, sample]
  76
  77   bwamem-gatk-report:
  78     run: ./helper/bwamem-gatk-report-wf.cwl
  79     scatter: [fastq1, fastq2, sample]
  80     scatterMethod: dotproduct
  81     in:
  82       fastq1: getfastq/fastq1
  83       fastq2: getfastq/fastq2
  84       reference: reference
  85       fullintervallist: fullintervallist
  86       sample: getfastq/sample
  87       knownsites1: knownsites1
  88       knownsites2: knownsites2
  89       scattercount: scattercount
  90       clinvarvcf: clinvarvcf
  91       reportfunc: reportfunc
  92       headhtml: headhtml
  93       tailhtml: tailhtml
  94     out: [qc-html,qc-zip,gvcf,report]
  95
  96 s:codeRepository: https://github.com/arvados/arvados-tutorial
  97 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
  98
  99 $namespaces:
 100  s: https://schema.org/
 101  edam: http://edamontology.org/
 102
 103 #$schemas:
 104 # - https://schema.org/version/latest/schema.rdf
 105 # - http://edamontology.org/EDAM_1.18.owl