somatic_SNVs_pipeline/workflow.wdl

#import "./tasks/fastp.wdl" as fastp
import "./tasks/BQSR.wdl" as Sentieon_BQSR
import "./tasks/TNseq.wdl" as TNseq
import "./tasks/TNscope.wdl" as TNscope
import "./tasks/manta.wdl" as manta_calling
import "./tasks/strelka.wdl" as strelka_calling
import "./tasks/bcftools.wdl" as bcftools
import "./tasks/bcftools_concat.wdl" as bcftools_concat
import "./tasks/annovar.wdl" as annovar


#如果是SEQC的结果，应该是'T_1'和'N_1'
workflow {{ project_name }} {
	#sample info
    File? normal_deduped_bam
    File? normal_deduped_bam_bai
	File? tumor_deduped_bam
	File? tumor_deduped_bam_bai
    String sample_id
	String Seq_platform
	#reference fasta
	String ref_fasta
	File ref_fasta_dir
	File annot_gff
    File dbsnp_dir
    File dbmills_dir
    String dbsnp
    String db_mills
	File germline_resource
	File germline_resource_tbi
	File annovar_database

	#docker
    String docker_fastp
	String docker_sentieon
	String SENTIEON_LICENSE
	String docker_qualimap
	String docker_manta
	String docker_strelka
	String docker_bcftools
	String docker_annovar

	#config
    String cluster_config
    String disk_size
	Boolean? fastqc
	Boolean bamqc
	Boolean Annovar


if (normal_deduped_bam != ""){

		call Sentieon_BQSR.Sentieon_BQSR as Sentieon_BQSR_normal{
			input:
				ref_dir=ref_fasta_dir,
				dbsnp_dir=dbsnp_dir,
				dbmills_dir = dbmills_dir,
				sample_id=sample_id+'_N',
				ref_fasta=ref_fasta,
				dbsnp=dbsnp,
				db_mills=db_mills,
				deduped_bam=normal_deduped_bam,
				deduped_bam_index=normal_deduped_bam_bai,
				docker=docker_sentieon,
				cluster_config=cluster_config,
				disk_size=disk_size,
				SENTIEON_LICENSE=SENTIEON_LICENSE
		}

	}

if (tumor_deduped_bam != ""){

		call Sentieon_BQSR.Sentieon_BQSR as Sentieon_BQSR_tumor{
			input:
				ref_dir=ref_fasta_dir,
				dbsnp_dir=dbsnp_dir,
				dbmills_dir = dbmills_dir,
				sample_id=sample_id+'_T',
				ref_fasta=ref_fasta,
				dbsnp=dbsnp,
				db_mills=db_mills,
				deduped_bam=tumor_deduped_bam,
				deduped_bam_index=tumor_deduped_bam_bai,
				docker=docker_sentieon,
				cluster_config=cluster_config,
				disk_size=disk_size,
				SENTIEON_LICENSE=SENTIEON_LICENSE
		}

	}

call TNseq.sentieon_TNseq as sentieon_TNseq{
	input:
		sample_id = sample_id,
		tumor_bam = Sentieon_BQSR_tumor.recaled_bam,
		tumor_bam_bai = Sentieon_BQSR_tumor.recaled_bam_index,
		normal_bam = Sentieon_BQSR_normal.recaled_bam,
		normal_bam_bai = Sentieon_BQSR_normal.recaled_bam_index,
		tumor_name = sample_id+'_T_1',
		normal_name = sample_id+'_N_1',

		ref_dir = ref_fasta_dir,
		ref_fasta = ref_fasta,
		germline_resource = germline_resource,
		germline_resource_tbi = germline_resource_tbi,

		docker = docker_sentieon,
		cluster_config = cluster_config,
		disk_size = disk_size,
		SENTIEON_LICENSE = SENTIEON_LICENSE
}

call bcftools.bcftools as TNseq_bcftools {
	input:
	vcf=sentieon_TNseq.vcf,
	fasta=ref_fasta,
	ref_dir=ref_fasta_dir,
	docker=docker_bcftools,
	cluster_config=cluster_config,
	disk_size=disk_size
}

call TNscope.sentieon_TNscope as sentieon_TNscope{
	input:
		sample_id = sample_id,
		tumor_bam = tumor_deduped_bam,
		tumor_bam_bai = tumor_deduped_bam_bai,
		normal_bam = normal_deduped_bam,
		normal_bam_bai = normal_deduped_bam_bai,
		tumor_name = sample_id+'_T_1',
		normal_name = sample_id+'_N_1',
		tumor_recall_data = Sentieon_BQSR_tumor.recal_table,
		normal_recall_data = Sentieon_BQSR_normal.recal_table,

		ref_dir = ref_fasta_dir,
		ref_fasta = ref_fasta,
		dbsnp_dir = dbsnp_dir,
		dbsnp = dbsnp,
		# excute env
		docker = docker_sentieon,
		cluster_config = cluster_config,
		disk_size = disk_size,
		SENTIEON_LICENSE = SENTIEON_LICENSE
}

call bcftools.bcftools as TNscope_bcftools {
	input:
	vcf=sentieon_TNscope.vcf,
	fasta=ref_fasta,
	ref_dir=ref_fasta_dir,
	docker=docker_bcftools,
	cluster_config=cluster_config,
	disk_size=disk_size
}

call manta_calling.manta_calling as manta_calling{
	input:
		tumor_bam = Sentieon_BQSR_tumor.recaled_bam,
		tumor_bam_bai = Sentieon_BQSR_tumor.recaled_bam_index,
		normal_bam = Sentieon_BQSR_normal.recaled_bam,
		normal_bam_bai = Sentieon_BQSR_normal.recaled_bam_index,
		ref_fasta = ref_fasta,
		ref_dir = ref_fasta_dir,
		sample_id = sample_id,
		docker = docker_manta,
		cluster_config = cluster_config,
		disk_size = disk_size

}

call strelka_calling.strelka_calling as strelka_calling{
	input:
		tumor_bam = Sentieon_BQSR_tumor.recaled_bam,
		tumor_bam_bai = Sentieon_BQSR_tumor.recaled_bam_index,
		normal_bam = Sentieon_BQSR_normal.recaled_bam,
		normal_bam_bai = Sentieon_BQSR_normal.recaled_bam_index,
		ref_fasta = ref_fasta,
		ref_dir = ref_fasta_dir,
		sample_id = sample_id,
		manta_indel_vcf = manta_calling.manta_indel_vcf,
		manta_indel_vcf_index = manta_calling.manta_indel_vcf_index,
		
		docker=docker_strelka,
		cluster_config=cluster_config,
		disk_size=disk_size
}

call bcftools_concat.bcftools as bcftools_concat{
        input:
            ref_dir=ref_fasta_dir,
            fasta=ref_fasta,
            vcf_indels=strelka_calling.indel_vcf,
            vcf_snvs=strelka_calling.snv_vcf,
            sample_id=sample_id,
            docker=docker_bcftools,
            cluster_config=cluster_config,
            disk_size=disk_size

    }

call bcftools.bcftools as strelka_bcftools {
	input:
	vcf=bcftools_concat.concat_vcf,
	fasta=ref_fasta,
	ref_dir=ref_fasta_dir,
	docker=docker_bcftools,
	cluster_config=cluster_config,
	disk_size=disk_size
}

if (Annovar){
	call annovar.ANNOVAR as TNseq_ANNOVAR {
		input:
			vcf=TNseq_bcftools.norm_vcf,
			annovar_database=annovar_database,
			docker=docker_annovar,
			cluster_config=cluster_config,
			disk_size=disk_size
  }

  	call annovar.ANNOVAR as TNscope_ANNOVAR {
		input:
			vcf=TNscope_bcftools.norm_vcf,
			annovar_database=annovar_database,
			docker=docker_annovar,
			cluster_config=cluster_config,
			disk_size=disk_size
  }

    call annovar.ANNOVAR as strelka_ANNOVAR {
		input:
			vcf=strelka_bcftools.norm_vcf,
			annovar_database=annovar_database,
			docker=docker_annovar,
			cluster_config=cluster_config,
			disk_size=disk_size
  }
}

}