WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl

   1 cwlVersion: v1.1
   2 class: Workflow
   3 label: WGS processing workflow for single sample
   4
   5 requirements:
   6   SubworkflowFeatureRequirement: {}
   7
   8 inputs:
   9   fastq1:
  10     type: File
  11     format: edam:format_1930 # FASTQ
  12     label: One of set of pair-end FASTQs (R1)
  13   fastq2:
  14     type: File
  15     format: edam:format_1930 # FASTQ
  16     label: One of set of pair-end FASTQs (R2)
  17   reference:
  18     type: File
  19     format: edam:format_1929 # FASTA
  20     label: Reference genome
  21     secondaryFiles:
  22       - .amb
  23       - .ann
  24       - .bwt
  25       - .pac
  26       - .sa
  27       - .fai
  28       - ^.dict
  29   fullintervallist:
  30     type: File
  31     label: Full list of intervals to operate over
  32   sample:
  33     type: string
  34     label: Sample Name
  35   knownsites1:
  36     type: File
  37     format: edam:format_3016 # VCF
  38     label: VCF of known SNPS sites for BQSR
  39     secondaryFiles:
  40       - .idx
  41   knownsites2:
  42     type: File
  43     format: edam:format_3016 # VCF
  44     label: VCF of known indel sites for BQSR
  45     secondaryFiles:
  46       - .tbi
  47   scattercount:
  48     type: string
  49     label: Desired split for variant calling
  50   clinvarvcf:
  51     type: File
  52     format: edam:format_3016 # VCF
  53     label: Reference VCF for ClinVar
  54   reportfunc:
  55     type: File
  56     label: Function used to create HTML report
  57   headhtml:
  58     type: File
  59     format: edam:format_2331 # HTML
  60     label: Header for HTML report
  61   tailhtml:
  62     type: File
  63     format: edam:format_2331 # HTML
  64     label: Footer for HTML report
  65
  66 outputs:
  67   qc-html:
  68     type: File[]
  69     label: FASTQ QC reports
  70     format: edam:format_2331 # HTML
  71     outputSource: fastqc/out-html
  72   qc-zip:
  73     type: File[]
  74     label: Zip files of FASTQ QC report and associated data
  75     outputSource: fastqc/out-zip
  76   gvcf:
  77     type: File
  78     outputSource: haplotypecaller/gatheredgvcf
  79     format: edam:format_3016 # GVCF
  80     label: GVCF generated from GATK Haplotype Caller
  81   report:
  82     type: File
  83     outputSource: generate-report/report
  84     format: edam:format_2331 # HTML
  85     label: ClinVar variant report
  86
  87 steps:
  88   fastqc:
  89     run: fastqc.cwl
  90     in:
  91       fastq1: fastq1
  92       fastq2: fastq2
  93     out: [out-html, out-zip]
  94   bwamem-samtools-view:
  95     run: bwamem-samtools-view.cwl
  96     in:
  97       fastq1: fastq1
  98       fastq2: fastq2
  99       reference: reference
 100       sample: sample
 101     out: [bam]
 102   samtools-sort:
 103     run: samtools-sort.cwl
 104     in:
 105       bam: bwamem-samtools-view/bam
 106       sample: sample
 107     out: [sortedbam]
 108   mark-duplicates:
 109     run: mark-duplicates.cwl
 110     in:
 111       bam: samtools-sort/sortedbam
 112     out: [dupbam,dupmetrics]
 113   samtools-index:
 114     run: samtools-index.cwl
 115     in:
 116       bam: mark-duplicates/dupbam
 117     out: [indexedbam]
 118   haplotypecaller:
 119     run: scatter-gatk-wf-with-interval.cwl
 120     in:
 121       reference: reference
 122       fullintervallist: fullintervallist
 123       bam: samtools-index/indexedbam
 124       sample: sample
 125       scattercount: scattercount
 126       knownsites1: knownsites1
 127       knownsites2: knownsites2
 128     out: [gatheredgvcf]
 129   generate-report:
 130     run: report-wf.cwl
 131     in:
 132       gvcf: haplotypecaller/gatheredgvcf
 133       sample: sample
 134       clinvarvcf: clinvarvcf
 135       reportfunc: reportfunc
 136       headhtml: headhtml
 137       tailhtml: tailhtml
 138     out: [report]
 139
 140 s:codeRepository: https://github.com/arvados/arvados-tutorial
 141 s:license: https://www.gnu.org/licenses/agpl-3.0.en.html
 142
 143 $namespaces:
 144  s: https://schema.org/
 145  edam: http://edamontology.org/
 146
 147 #$schemas:
 148 # - https://schema.org/version/latest/schema.rdf
 149 # - http://edamontology.org/EDAM_1.18.owl