variant-calling/workflow.wdl

import "./tasks/mapping.wdl" as mapping
import "./tasks/Metrics.wdl" as Metrics
import "./tasks/Dedup.wdl" as Dedup
import "./tasks/deduped_Metrics.wdl" as deduped_Metrics
import "./tasks/BQSR.wdl" as BQSR
import "./tasks/Haplotyper.wdl" as Haplotyper
import "./tasks/TNseq.wdl" as TNseq
import "./tasks/TNscope.wdl" as TNscope
import "./tasks/somatic.wdl" as somatic
import "./tasks/processSomatic.wdl" as processSomatic
import "./tasks/somaticFilter.wdl" as somaticFilter
import "./tasks/replaceRG.wdl" as replaceRG
import "./tasks/pindel.wdl" as pindel
import "./tasks/ANNOVAR.wdl" as ANNOVAR
import "./tasks/VEP.wdl" as VEP


workflow {{ project_name }} {
  
  String sample_id
  File? tumor_fastq_1
  File? tumor_fastq_2
  File? normal_fastq_1
  File? normal_fastq_2
  File? tumor_deduped_bam
  File? normal_deduped_bam

  Boolean input_fastq
  Boolean input_bam

  String SENTIEON_INSTALL_DIR
  String SENTIEON_LICENSE
  String sentieon_docker
  String varscan_docker
  String pindel_docker
  String samtools_docker
  String annovar_docker
  String vep_docker

  String pl
  File ref_dir
  String fasta
  File dbmills_dir
  String db_mills
  File dbsnp_dir
  String dbsnp
  File germline_resource
  File germline_resource_tbi
  String hg
  File annovar_database
  
  File? regions
  Int? interval_padding
  File? tnseq_pon
  File? tnscope_pon

  String vep_path
  File cache
  String species
  String vcf2maf_path

  String disk_size
  String cluster_config

  Boolean haplotyper
  Boolean tnseq
  Boolean tnscope
  Boolean varscan

  Boolean annovar
  Boolean vep
  Boolean only_pass

  if (input_fastq) {
    if (tumor_fastq_1!= "") {
      call mapping.mapping as tumor_mapping {
        input: 
        group=sample_id + '.T',
        sample=sample_id + '.T',
        fastq_1=tumor_fastq_1,
        fastq_2=tumor_fastq_2,
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        pl=pl,
        fasta=fasta,
        ref_dir=ref_dir,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call Metrics.Metrics as tumor_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        sorted_bam=tumor_mapping.sorted_bam,
        sorted_bam_index=tumor_mapping.sorted_bam_index,
        sample=sample_id + '.T',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call Dedup.Dedup as tumor_Dedup {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sorted_bam=tumor_mapping.sorted_bam,
        sorted_bam_index=tumor_mapping.sorted_bam_index,
        sample=sample_id + '.T',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call deduped_Metrics.deduped_Metrics as tumor_deduped_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=tumor_Dedup.deduped_bam,
        deduped_bam_index=tumor_Dedup.deduped_bam_index,
        sample=sample_id + '.T',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call BQSR.BQSR as tumor_BQSR {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=tumor_Dedup.deduped_bam,
        deduped_bam_index=tumor_Dedup.deduped_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '.T',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }
    
    if (normal_fastq_1!= "") {
      call mapping.mapping as normal_mapping {
        input: 
        group=sample_id + '.N',
        sample=sample_id + '.N',
        fastq_1=normal_fastq_1,
        fastq_2=normal_fastq_2,
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        pl=pl,
        fasta=fasta,
        ref_dir=ref_dir,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call Metrics.Metrics as normal_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        sorted_bam=normal_mapping.sorted_bam,
        sorted_bam_index=normal_mapping.sorted_bam_index,
        sample=sample_id + '.N',
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call Dedup.Dedup as normal_Dedup {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sorted_bam=normal_mapping.sorted_bam,
        sorted_bam_index=normal_mapping.sorted_bam_index,
        sample=sample_id + '.N',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call deduped_Metrics.deduped_Metrics as normal_deduped_Metrics {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=normal_Dedup.deduped_bam,
        deduped_bam_index=normal_Dedup.deduped_bam_index,
        sample=sample_id + '.N',
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call BQSR.BQSR as normal_BQSR {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=normal_Dedup.deduped_bam,
        deduped_bam_index=normal_Dedup.deduped_bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '.N',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }

    if (haplotyper) {
      call Haplotyper.Haplotyper as Haplotyper {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id + '.N',
        fasta=fasta,
        ref_dir=ref_dir,
        recaled_bam=normal_BQSR.recaled_bam,
        recaled_bam_index=normal_BQSR.recaled_bam_index,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }
     
    if (pindel) {
      call pindel.pindel as pindel {
        input:	
        sample_id=sample_id + '.N',
        bam=normal_BQSR.recaled_bam,
        bam_index=normal_BQSR.recaled_bam_index,
        fasta=fasta,
        ref_dir=ref_dir,
        docker=pindel_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }

    if (tnseq) {
      call TNseq.TNseq as TNseq {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id,
        normal_recaled_bam=normal_BQSR.recaled_bam,
        normal_recaled_bam_index=normal_BQSR.recaled_bam_index,
        tumor_recaled_bam=tumor_BQSR.recaled_bam,
        tumor_recaled_bam_index=tumor_BQSR.recaled_bam_index,
        normal_name=sample_id + ".N",
        tumor_name=sample_id + ".T",
        fasta=fasta,
        ref_dir=ref_dir,
        regions=regions,
        interval_padding=interval_padding,
        germline_resource=germline_resource,
        germline_resource_tbi=germline_resource_tbi,
        pon_vcf=tnseq_pon,
        docker=sentieon_docker,
        cluster_config=cluster_config,
        disk_size=disk_size
      }

      if (annovar) {
        call ANNOVAR.ANNOVAR as TNseq_annovar {
          input:
          vcf=TNseq.TNseq_filter_vcf,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as TNseq_VEP {
          input: 
          vcf=TNseq.TNseq_filter_vcf,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }

    if (tnscope) {
      call TNscope.TNscope as TNscope {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id,
        normal_recaled_bam=normal_BQSR.recaled_bam,
        normal_recaled_bam_index=normal_BQSR.recaled_bam_index,
        tumor_recaled_bam=tumor_BQSR.recaled_bam,
        tumor_recaled_bam_index=tumor_BQSR.recaled_bam_index,
        normal_name=sample_id + ".N",
        tumor_name=sample_id + ".T",
        fasta=fasta,
        ref_dir=ref_dir,
        regions=regions,
        interval_padding=interval_padding,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        pon_vcf=tnscope_pon,
        docker=sentieon_docker,
        cluster_config=cluster_config,
        disk_size=disk_size
      }
      
      if (annovar) {
        call ANNOVAR.ANNOVAR as TNscope_annovar {
          input:
          vcf=TNscope.TNscope_vcf,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as TNscope_VEP {
          input: 
          vcf=TNscope.TNscope_vcf,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }

    if (varscan) {
      call somatic.somatic as somatic {
        input:
        sample=sample_id,
        normal_bam=normal_BQSR.recaled_bam,
        normal_bam_index=normal_BQSR.recaled_bam_index,
        tumor_bam=tumor_BQSR.recaled_bam,
        tumor_bam_index=tumor_BQSR.recaled_bam_index,
        ref_dir=ref_dir,
        fasta=fasta,    
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call processSomatic.processSomatic as processSomatic {
        input:
        sample=sample_id,
        varscan_snp=somatic.varscan_snp,
        varscan_indel=somatic.varscan_indel,   
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call somaticFilter.somaticFilter as somaticFilter {
        input:
        sample=sample_id,
        varscan_snp_somatic_hc=processSomatic.varscan_snp_somatic_hc,
        varscan_snp_germline_hc=processSomatic.varscan_snp_germline_hc,
        varscan_snp_loh_hc=processSomatic.varscan_snp_loh_hc,
        varscan_indel_somatic_hc=processSomatic.varscan_indel_somatic_hc,
        varscan_indel_germline_hc=processSomatic.varscan_indel_germline_hc,
        varscan_indel_loh_hc=processSomatic.varscan_indel_loh_hc,
        varscan_indel=somatic.varscan_indel,   
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      if (annovar) {
        call ANNOVAR.ANNOVAR as VarScan_annovar {
          input:
          vcf=somaticFilter.varscan_somatic_filter,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as VarScan_VEP {
          input: 
          vcf=somaticFilter.varscan_somatic_filter,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }
  }

  if (input_bam) {
    if (tumor_deduped_bam != "") {
      call replaceRG.replaceRG as tumor_replaceRG {
        input:
        bam=tumor_deduped_bam,
        group=group_id + '.T',
        sample=sample_id + '.T',
        pl=pl,
        docker=samtools_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call BQSR.BQSR as tumor_BQSR_fb {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=tumor_replaceRG.bam,
        deduped_bam_index=tumor_replaceRG.bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '.T',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }
    
    if (normal_deduped_bam != "") {
      call replaceRG.replaceRG as normal_replaceRG {
        input:
        bam=normal_deduped_bam,
        group=group_id + '.N',
        sample=sample_id + '.N',
        pl=pl,
        docker=samtools_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
      
      call BQSR.BQSR as normal_BQSR_fb {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        fasta=fasta,
        ref_dir=ref_dir,
        deduped_bam=normal_replaceRG.bam,
        deduped_bam_index=normal_replaceRG.bam_index,
        db_mills=db_mills,
        dbmills_dir=dbmills_dir,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        sample=sample_id + '.N',
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }
    
    if (haplotyper) {
      call Haplotyper.Haplotyper as Haplotyper_fb {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id + '.N',
        fasta=fasta,
        ref_dir=ref_dir,
        recaled_bam=normal_BQSR_fb.recaled_bam,
        recaled_bam_index=normal_BQSR_fb.recaled_bam_index,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        regions=regions,
        docker=sentieon_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }

    if (pindel) {
      call pindel.pindel as pindel_fb {
        input:	
        sample_id=sample_id + '.N',
        bam=normal_BQSR_fb.recaled_bam,
        bam_index=normal_BQSR_fb.recaled_bam_index,
        fasta=fasta,
        ref_dir=ref_dir,
        docker=pindel_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }
    }

    if (tnseq) {
      call TNseq.TNseq as TNseq_fb {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id,
        normal_recaled_bam=normal_BQSR_fb.recaled_bam,
        normal_recaled_bam_index=normal_BQSR_fb.recaled_bam_index,
        tumor_recaled_bam=tumor_BQSR_fb.recaled_bam,
        tumor_recaled_bam_index=tumor_BQSR_fb.recaled_bam_index,
        normal_name=sample_id + ".N",
        tumor_name=sample_id + ".T",
        fasta=fasta,
        ref_dir=ref_dir,
        regions=regions,
        interval_padding=interval_padding,
        germline_resource=germline_resource,
        germline_resource_tbi=germline_resource_tbi,
        pon_vcf=tnseq_pon,
        docker=sentieon_docker,
        cluster_config=cluster_config,
        disk_size=disk_size
      }

      if (annovar) {
        call ANNOVAR.ANNOVAR as TNseq_annovar_fb {
          input:
          vcf=TNseq_fb.TNseq_filter_vcf,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as TNseq_VEP_fb {
          input: 
          vcf=TNseq_fb.TNseq_filter_vcf,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }

    if (tnscope) {
      call TNscope.TNscope as TNscope_fb {
        input:
        SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
        SENTIEON_LICENSE=SENTIEON_LICENSE,
        sample=sample_id,
        normal_recaled_bam=normal_BQSR_fb.recaled_bam,
        normal_recaled_bam_index=normal_BQSR_fb.recaled_bam_index,
        tumor_recaled_bam=tumor_BQSR_fb.recaled_bam,
        tumor_recaled_bam_index=tumor_BQSR_fb.recaled_bam_index,
        normal_name=sample_id + ".N",
        tumor_name=sample_id + ".T",
        fasta=fasta,
        ref_dir=ref_dir,
        regions=regions,
        interval_padding=interval_padding,
        dbsnp=dbsnp,
        dbsnp_dir=dbsnp_dir,
        pon_vcf=tnscope_pon,
        docker=sentieon_docker,
        cluster_config=cluster_config,
        disk_size=disk_size
      }
      
      if (annovar) {
        call ANNOVAR.ANNOVAR as TNscope_annovar_fb {
          input:
          vcf=TNscope_fb.TNscope_vcf,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as TNscope_VEP_fb {
          input: 
          vcf=TNscope_fb.TNscope_vcf,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }

    if (varscan) {
      call somatic.somatic as somatic_fb {
        input:
        sample=sample_id,
        normal_bam=normal_BQSR_fb.recaled_bam,
        normal_bam_index=normal_BQSR_fb.recaled_bam_index,
        tumor_bam=tumor_BQSR_fb.recaled_bam,
        tumor_bam_index=tumor_BQSR_fb.recaled_bam_index,
        ref_dir=ref_dir,
        fasta=fasta,    
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call processSomatic.processSomatic as processSomatic_fb {
        input:
        sample=sample_id,
        varscan_snp=somatic_fb.varscan_snp,
        varscan_indel=somatic_fb.varscan_indel,   
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      call somaticFilter.somaticFilter as somaticFilter_fb {
        input:
        sample=sample_id,
        varscan_snp_somatic_hc=processSomatic_fb.varscan_snp_somatic_hc,
        varscan_snp_germline_hc=processSomatic_fb.varscan_snp_germline_hc,
        varscan_snp_loh_hc=processSomatic_fb.varscan_snp_loh_hc,
        varscan_indel_somatic_hc=processSomatic_fb.varscan_indel_somatic_hc,
        varscan_indel_germline_hc=processSomatic_fb.varscan_indel_germline_hc,
        varscan_indel_loh_hc=processSomatic_fb.varscan_indel_loh_hc,
        varscan_indel=somatic_fb.varscan_indel,   
        docker=varscan_docker,
        disk_size=disk_size,
        cluster_config=cluster_config
      }

      if (annovar) {
        call ANNOVAR.ANNOVAR as VarScan_annovar_fb {
          input:
          vcf=somaticFilter_fb.varscan_somatic_filter,
          hg=hg,
          only_pass=only_pass,
          annovar_database=annovar_database,
          docker=annovar_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }

      if (vep) {
        call VEP.VEP as VarScan_VEP_fb {
          input: 
          vcf=somaticFilter_fb.varscan_somatic_filter,
          hg=hg,
          only_pass=only_pass,
          sample_id=sample_id,
          tumor_id=sample_id + ".T",
          normal_id=sample_id + ".N",
          ref_dir=ref_dir,
          fasta=fasta,
          vep_path=vep_path,
          cache=cache,
          species=species,
          vcf2maf_path=vcf2maf_path,
          docker=vep_docker,
          cluster_config=cluster_config,
          disk_size=disk_size
        }
      }
    }
  }
}