first commit

README.md

@@ -0,0 +1,8 @@
+#### Usage
+$ open-choppy-env
+$ choppy install ...
+
+#### inputSamplesFile
+sample_id	tumor_fastq_1	tumor_fastq_2	normal_fastq_1	normal_fastq_1
+
+inputSamplesFile is in the form of tsv, and should be put into Aliyun OSS.

defaults

@@ -0,0 +1,23 @@
+{
+    "fasta": "hg19_nochr.fa",
+    "ref_dir": "oss://pgx-reference-data/hg19/",
+    "dbsnp": "dbsnp_138.hg19_nochr_sorted.vcf",
+    "dbsnp_dir": "oss://pgx-reference-data/hg19/",
+    "SENTIEON_INSTALL_DIR": "/opt/sentieon-genomics",
+    "SENTIEON_LICENSE": "192.168.0.55:8990",
+    "dbmills_dir": "oss://pgx-reference-data/hg19/",
+    "db_mills": "Mills_and_1000G_gold_standard.indels.hg19_nochr.vcf",
+    "sentieon_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/sentieon-genomics:v2017.11.04",
+    "annovar_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/annovar:v2018.04",
+    "vcf2maf_r_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/r-base:4.0.2",
+    "bcftools_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/bcftools:v1.9",
+    "database": "oss://pgx-reference-data/annovar/",
+    "regions": "oss://pgx-reference-data/reference/wes_bedfiles/agilent_v6/agilent_v6.bed",
+    "PONmode": true,
+    "cosmic_vcf": "CosmicCodingMuts.hg19.v91.vcf",
+    "cosmic_dir": "oss://pgx-reference-data/reference/cosmic/",
+    "disk_size": "200",
+    "cluster_config": "OnDemand bcs.a2.3xlarge img-ubuntu-vpc",
+    "set_annovar": true,
+    "set_vcf2maf": true
+}

inputs

@@ -0,0 +1,25 @@
+{
+  "{{ project_name }}.fasta": "{{ fasta }}",
+  "{{ project_name }}.ref_dir": "{{ ref_dir }}",
+  "{{ project_name }}.dbsnp": "{{ dbsnp }}",
+  "{{ project_name }}.dbsnp_dir": "{{ dbsnp_dir }}",
+  "{{ project_name }}.SENTIEON_INSTALL_DIR": "{{ SENTIEON_INSTALL_DIR }}",
+  "{{ project_name }}.SENTIEON_LICENSE": "{{ SENTIEON_LICENSE }}",
+  "{{ project_name }}.dbmills_dir": "{{ dbmills_dir }}",
+  "{{ project_name }}.db_mills": "{{ db_mills }}",
+  "{{ project_name }}.sentieon_docker": "{{ sentieon_docker }}",
+  "{{ project_name }}.annovar_docker": "{{ annovar_docker }}",
+  "{{ project_name }}.vcf2maf_r_docker": "{{ vcf2maf_r_docker }}",
+  "{{ project_name }}.bcftools_docker": "{{ bcftools_docker }}",
+  "{{ project_name }}.database": "{{ database }}",
+  "{{ project_name }}.inputSamplesFile": "{{ inputSamplesFile }}",
+  "{{ project_name }}.inputponfile": "{{ inputponfile }}",
+  "{{ project_name }}.regions": "{{ regions }}",
+  "{{ project_name }}.PONmode": "{{ PONmode }}",
+  "{{ project_name }}.cosmic_vcf": "{{ cosmic_vcf }}",
+  "{{ project_name }}.cosmic_dir": "{{ cosmic_dir }}",
+  "{{ project_name }}.disk_size": "{{ disk_size }}",
+  "{{ project_name }}.cluster_config": "{{ cluster_config }}",
+  "{{ project_name }}.set_annovar": "{{ set_annovar }}",
+  "{{ project_name }}.set_vcf2maf": "{{ set_vcf2maf }}"
+}

tasks/BQSR.wdl

@@ -0,0 +1,52 @@
+task BQSR {
+	
+	File ref_dir
+	File dbsnp_dir
+	File dbmills_dir
+	
+	String sample
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String fasta
+	File regions
+	String dbsnp
+	String db_mills
+	File realigned_bam1
+	File realigned_bam_index1
+	String docker
+	String cluster_config
+	String disk_size
+
+	
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam1} --interval ${regions} --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table
+
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${realigned_bam1} -q ${sample}_recal_data.table --algo QualCal -k ${dbsnp_dir}/${dbsnp} -k ${dbmills_dir}/${db_mills} ${sample}_recal_data.table.post --algo ReadWriter ${sample}.sorted.deduped.realigned.recaled.bam
+
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt --algo QualCal --plot --before ${sample}_recal_data.table --after ${sample}_recal_data.table.post ${sample}_recal_data.csv
+
+		${SENTIEON_INSTALL_DIR}/bin/sentieon plot QualCal -o ${sample}_bqsrreport.pdf ${sample}_recal_data.csv
+
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"	
+	}
+
+	output {
+		File recal_table = "${sample}_recal_data.table"
+		File recal_post = "${sample}_recal_data.table.post"
+		File recaled_bam = "${sample}.sorted.deduped.realigned.recaled.bam"
+		File recaled_bam_index = "${sample}.sorted.deduped.realigned.recaled.bam.bai"
+		File recal_csv = "${sample}_recal_data.csv"
+		File bqsrreport_pdf = "${sample}_bqsrreport.pdf"
+	}
+}

tasks/Dedup.wdl

@@ -0,0 +1,43 @@
+task Dedup {
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+
+	File sorted_bam
+	File sorted_bam_index
+	String docker
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo LocusCollector --fun score_info ${sample}_score.txt
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -i ${sorted_bam} --algo Dedup --rmdup --score_info ${sample}_score.txt --metrics ${sample}_dedup_metrics.txt ${sample}.sorted.deduped.bam
+
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File score = "${sample}_score.txt"
+		File dedup_metrics = "${sample}_dedup_metrics.txt"
+		File Dedup_bam = "${sample}.sorted.deduped.bam"
+		File Dedup_bam_index = "${sample}.sorted.deduped.bam.bai"
+	}
+}
+
+
+
+
+
+

tasks/Haplotyper.wdl

@@ -0,0 +1,39 @@
+task Haplotyper {
+	
+	File ref_dir
+	File dbsnp_dir
+	
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String fasta
+	File recaled_bam
+	File recaled_bam_index
+	File regions
+	String dbsnp
+	String sample
+	String docker
+	String cluster_config
+	String disk_size
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)	
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver --interval ${regions} -r ${ref_dir}/${fasta} -t $nt -i ${recaled_bam} --algo Haplotyper -d ${dbsnp_dir}/${dbsnp} ${sample}_hc.vcf
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File vcf = "${sample}_hc.vcf"
+		File vcf_idx = "${sample}_hc.vcf.idx"
+	}
+}
+
+

tasks/Metrics.wdl

@@ -0,0 +1,57 @@
+task Metrics {
+
+
+	File ref_dir
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+	String docker
+	String cluster_config
+
+	String fasta
+	File sorted_bam
+	File sorted_bam_index
+	String disk_size
+	
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${sorted_bam} --algo MeanQualityByCycle ${sample}_mq_metrics.txt --algo QualDistribution ${sample}_qd_metrics.txt --algo GCBias --summary ${sample}_gc_summary.txt ${sample}_gc_metrics.txt --algo AlignmentStat ${sample}_aln_metrics.txt --algo InsertSizeMetricAlgo ${sample}_is_metrics.txt --algo CoverageMetrics --omit_base_output ${sample}_coverage_metrics
+
+		${SENTIEON_INSTALL_DIR}/bin/sentieon plot metrics -o ${sample}_metrics_report.pdf gc=${sample}_gc_metrics.txt qd=${sample}_qd_metrics.txt mq=${sample}_mq_metrics.txt isize=${sample}_is_metrics.txt
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File qd_metrics = "${sample}_qd_metrics.txt"
+		File qd_metrics_pdf = "${sample}_qd_metrics.pdf"
+		File mq_metrics = "${sample}_mq_metrics.txt"
+		File mq_metrics_pdf = "${sample}_mq_metrics.pdf"
+		File is_metrics = "${sample}_is_metrics.txt"
+		File is_metrics_pdf = "${sample}_is_metrics.pdf"
+		File gc_summary = "${sample}_gc_summary.txt"
+		File gc_metrics = "${sample}_gc_metrics.txt"
+		File gc_metrics_pdf = "${sample}_gc_metrics.pdf"
+		File aln_metrics = "${sample}_aln_metrics.txt"
+		File coverage_metrics_sample_summary = "${sample}_coverage_metrics.sample_summary"
+		File coverage_metrics_sample_statistics = "${sample}_coverage_metrics.sample_statistics"
+		File coverage_metrics_sample_interval_statistics = "${sample}_coverage_metrics.sample_interval_statistics"
+		File coverage_metrics_sample_cumulative_coverage_proportions = "${sample}_coverage_metrics.sample_cumulative_coverage_proportions"
+		File coverage_metrics_sample_cumulative_coverage_counts = "${sample}_coverage_metrics.sample_cumulative_coverage_counts"
+	}
+
+}
+
+
+
+
+

tasks/PON.wdl

@@ -0,0 +1,46 @@
+task PON {
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+	Boolean PONmode
+
+	File ref_dir
+	String fasta
+	File cosmic_dir
+	String cosmic_vcf
+	File dbsnp_dir
+	String dbsnp
+
+	File normal_recaled_bam
+	File normal_recaled_bam_index
+	String docker
+	String cluster_config
+	String disk_size
+	
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		if [ ${PONmode} ]; then
+			mkdir -p /cromwell_root/tmp/cosmic/
+			cp ${cosmic_dir}/${cosmic_vcf} /cromwell_root/tmp/cosmic/
+			${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/cosmic/${cosmic_vcf}
+			${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${normal_recaled_bam} --algo TNhaplotyper --detect_pon --cosmic /cromwell_root/tmp/cosmic/${cosmic_vcf} --dbsnp ${dbsnp_dir}/${dbsnp} ${sample}_pon.vcf
+		else
+			touch ${sample}_pon.vcf
+		fi
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"	
+	}
+
+	output {
+		File pon_vcf = "${sample}_pon.vcf"
+	}
+}

tasks/Realigner.wdl

@@ -0,0 +1,42 @@
+task Realigner {
+
+	File ref_dir
+	File dbmills_dir
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+	String fasta
+
+	File regions
+	File Dedup_bam
+	File Dedup_bam_index
+	String db_mills
+	String docker	
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${Dedup_bam} --algo Realigner -k ${dbmills_dir}/${db_mills} --interval_list ${regions} ${sample}.sorted.deduped.realigned.bam
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File realigner_bam = "${sample}.sorted.deduped.realigned.bam"
+		File realigner_bam_index = "${sample}.sorted.deduped.realigned.bam.bai"
+
+	}
+}
+
+

tasks/TNscope.wdl

@@ -0,0 +1,57 @@
+task TNscope {
+
+	File ref_dir
+	File dbsnp_dir
+	String sample
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String tumor_name
+	String normal_name
+	String docker
+	String cluster_config
+
+	String fasta
+	File corealigner_bam
+	File corealigner_bam_index
+	String dbsnp
+	String disk_size
+
+	Boolean PONmode
+	String? cosmic_vcf
+	File? cosmic_dir
+	File panel_of_normal_vcf
+	
+	
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+
+		if [ ${PONmode} ]; then
+			mkdir -p /cromwell_root/tmp/cosmic/
+			cp ${cosmic_dir}/${cosmic_vcf} /cromwell_root/tmp/cosmic/
+			${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/cosmic/${cosmic_vcf}
+			PONcommand="--pon ${panel_of_normal_vcf} --cosmic /cromwell_root/tmp/cosmic/${cosmic_vcf}"
+		else
+			PONcommand=""
+		fi
+		
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${corealigner_bam} --algo TNscope --tumor_sample ${tumor_name} --normal_sample ${normal_name} --dbsnp ${dbsnp_dir}/${dbsnp} PONcommand ${sample}.TNscope.TN.vcf
+		
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File TNscope_vcf= "${sample}.TNscope.TN.vcf"
+		File TNscope_vcf_index = "${sample}.TNscope.TN.vcf.idx"
+	}
+
+}

tasks/TNseq.wdl

@@ -0,0 +1,57 @@
+task TNseq {
+	File ref_dir
+	File dbsnp_dir
+	Array[String] sample
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String tumor_name
+	String normal_name
+	String docker
+	String cluster_config
+
+	String fasta
+	Array[File] corealigner_bam
+	Array[File] corealigner_bam_index
+	String dbsnp
+	String disk_size
+	
+	Boolean PONmode
+	String? cosmic_vcf
+	File? cosmic_dir
+	File panel_of_normal_vcf
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+
+		if [ ${PONmode} ]; then
+			mkdir -p /cromwell_root/tmp/cosmic/
+			cp ${cosmic_dir}/${cosmic_vcf} /cromwell_root/tmp/cosmic/
+			${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/cosmic/${cosmic_vcf}
+			PONcommand="--pon ${panel_of_normal_vcf} --cosmic /cromwell_root/tmp/cosmic/${cosmic_vcf}"
+		else
+			PONcommand=""
+		fi
+		
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${sep=" -i " corealigner_bam} --algo TNhaplotyper --tumor_sample ${tumor_name} --normal_sample ${normal_name} --dbsnp ${dbsnp_dir}/${dbsnp} PONcommand ${sep=" " sample}.TNseq.TN.vcf
+
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File TNseq_vcf= "${sep=" " sample}.TNseq.TN.vcf"
+		File TNseq_vcf_index = "${sep=" " sample}.TNseq.TN.vcf.idx"
+	}
+
+}
+
+

tasks/annovar.wdl

@@ -0,0 +1,35 @@
+task annovar {
+
+	File tnscope_vcf_file
+	File tnseq_vcf_file
+	File database
+	String sample
+	String docker
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		nt=$(nproc)
+		/installations/annovar/table_annovar.pl ${tnscope_vcf_file} ${database} -buildver hg19 -out ${sample}_tnscope -remove -protocol refGene,cytoBand,genomicSuperDups,snp138,ljb26_all,cosmic78,clinvar_20170130,popfreq_all_20150413,intervar_20170202 -operation g,r,r,f,f,f,f,f,f -nastring . -vcfinput -thread $nt
+		/installations/annovar/table_annovar.pl ${tnseq_vcf_file} ${database} -buildver hg19 -out ${sample}_tnseq -remove -protocol refGene,cytoBand,genomicSuperDups,snp138,ljb26_all,cosmic78,clinvar_20170130,popfreq_all_20150413,intervar_20170202 -operation g,r,r,f,f,f,f,f,f -nastring . -vcfinput -thread $nt
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File avinput_tnscope = "${sample}_tnscope.avinput"
+		File multianno_tnscope_txt = "${sample}_tnscope.hg19_multianno.txt"
+		File multianno_tnscope_vcf = "${sample}_tnscope.hg19_multianno.vcf"
+		File avinput_tnseq = "${sample}_tnseq.avinput"
+		File multianno_tnseq_txt = "${sample}_tnseq.hg19_multianno.txt"
+		File multianno_tnseq_vcf = "${sample}_tnseq.hg19_multianno.vcf"
+	}
+}

tasks/bcftools.wdl

@@ -0,0 +1,40 @@
+task bcftools {
+
+	Array[File] pon_vcf
+	String docker
+	String cluster_config
+	String disk_size
+	Boolean PONmode
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		
+		if [ ${PONmode} ]; then
+			mkdir -p /cromwell_root/tmp/bcftools/
+			for i in ${sep=" " pon_vcf}
+			do
+				bcftools view $i -Oz -o /cromwell_root/tmp/bcftools/$i.gz
+			done
+			
+			bcftools merge -m any -f PASS,. --force-samples /cromwell_root/tmp/bcftools/*.vcf.gz |\ 
+			bcftools plugin fill-AN-AC |\ 
+			bcftools filter -i 'SUM(AC)>1' > panel_of_normal.vcf
+		else
+			touch panel_of_normal.vcf
+		fi
+
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	
+	output {
+		File panel_of_normal_vcf = "panel_of_normal.vcf"
+	}
+}

tasks/corealigner.wdl

@@ -0,0 +1,46 @@
+task corealigner {
+	
+	File ref_dir
+	File dbsnp_dir
+	File dbmills_dir
+
+	Array[String] sample
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String docker
+	String cluster_config
+	String fasta
+
+	String dbsnp
+	String db_mills
+	Array[File] tumor_recaled_bam
+	Array[File] tumor_recaled_bam_index
+	Array[File] normal_recaled_bam
+	Array[File] normal_recaled_bam_index
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		#${sep=" " tumor_recaled_bam}
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${sep=" -i " tumor_recaled_bam} -i ${sep=" -i " normal_recaled_bam} --algo Realigner -k ${dbmills_dir}/${db_mills} -k ${dbsnp_dir}/${dbsnp} ${sep=" " sample}_corealigned.bam				
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File corealigner_bam = "${sep=" " sample}_corealigned.bam"
+		File corealigner_bam_index = "${sep=" " sample}_corealigned.bam.bai"
+	}	
+}
+
+
+

tasks/deduped_Metrics.wdl

@@ -0,0 +1,37 @@
+task deduped_Metrics {
+
+	File ref_dir
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+	String fasta
+	File Dedup_bam
+	File Dedup_bam_index
+	String docker
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${Dedup_bam} --algo CoverageMetrics --omit_base_output ${sample}_deduped_coverage_metrics --algo MeanQualityByCycle ${sample}_deduped_mq_metrics.txt --algo QualDistribution ${sample}_deduped_qd_metrics.txt --algo GCBias --summary ${sample}_deduped_gc_summary.txt ${sample}_deduped_gc_metrics.txt --algo AlignmentStat ${sample}_deduped_aln_metrics.txt --algo InsertSizeMetricAlgo ${sample}_deduped_is_metrics.txt
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/" 
+	}
+
+	output {
+		File deduped_coverage_metrics_sample_summary = "${sample}_deduped_coverage_metrics.sample_summary"
+		File deduped_coverage_metrics_sample_statistics = "${sample}_deduped_coverage_metrics.sample_statistics"
+		File deduped_coverage_metrics_sample_interval_statistics = "${sample}_deduped_coverage_metrics.sample_interval_statistics"
+		File deduped_coverage_metrics_sample_cumulative_coverage_proportions = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_proportions"
+		File deduped_coverage_metrics_sample_cumulative_coverage_counts = "${sample}_deduped_coverage_metrics.sample_cumulative_coverage_counts"
+	}
+}

tasks/mapping.wdl

@@ -0,0 +1,35 @@
+task mapping {
+
+	File ref_dir
+	String fasta
+	File fastq_1
+	File fastq_2
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String group
+	String sample
+	String pl
+	String docker
+	String cluster_config
+	String disk_size
+
+	command <<<
+		set -o pipefail
+		set -e	
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/bwa mem -M -R "@RG\tID:${group}\tSM:${sample}\tPL:${pl}" -t $nt -K 10000000 ${ref_dir}/${fasta} ${fastq_1} ${fastq_2} | ${SENTIEON_INSTALL_DIR}/bin/sentieon util sort -o ${sample}.sorted.bam -t $nt --sam2bam -i -
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File sorted_bam = "${sample}.sorted.bam"
+		File sorted_bam_index = "${sample}.sorted.bam.bai"
+	}
+}

tasks/vcf2maf.wdl

@@ -0,0 +1,26 @@
+task vcf2maf {
+
+	File? multianno_tnscope_txt
+	File? multianno_tnseq_txt
+	String sample
+	String docker
+	String cluster_config
+	String disk_size
+
+	command <<<
+		vcf2maf ${multianno_tnscope_txt} ${sample}_tnscope.MAF
+		vcf2maf ${multianno_tnseq_txt} ${sample}_tnseq.MAF
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File tnscope_maf = "${sample}_tnscope.MAF"
+		File tnseq_maf = "${sample}_tnseq.MAF"
+	}
+}

workflow.wdl

@@ -0,0 +1,145 @@
+
+import "./tasks/corealigner.wdl" as corealigner
+import "./tasks/bcftools.wdl" as bcftools
+import "./tasks/TNseq.wdl" as TNseq
+import "./tasks/TNscope.wdl" as TNscope
+import "./tasks/annovar.wdl" as annovar
+import "./tasks/vcf2maf.wdl" as vcf2maf
+
+
+workflow {{ project_name }} {
+
+	File inputSamplesFile
+	Array[Array[File]] inputSamples = read_tsv(inputSamplesFile)
+	#Array[String] sample
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sentieon_docker
+	String annovar_docker
+	String vcf2maf_r_docker
+	String bcftools_docker
+	
+	File ref_dir
+	String fasta
+	File dbmills_dir
+	String db_mills
+	File dbsnp_dir
+	String dbsnp
+	File regions
+	File database
+	String disk_size
+	String cluster_config
+
+	Boolean PONmode
+	File? cosmic_dir
+	String? cosmic_vcf
+
+	Boolean set_annovar
+	Boolean set_vcf2maf
+
+
+	File inputponfile
+	Array[Array[File]] bcf = read_tsv(inputponfile)
+
+	call bcftools.bcftools as bcftools {
+		input:
+		PONmode=PONmode,
+		pon_vcf=bcf,
+		docker=bcftools_docker,
+		disk_size=disk_size,
+		cluster_config=cluster_config
+	}
+	
+	
+	scatter (sample in inputSamples) {
+		call corealigner.corealigner as corealigner {
+	        input:
+    	    SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+    	    SENTIEON_LICENSE=SENTIEON_LICENSE,
+    	    fasta=fasta,
+    	    ref_dir=ref_dir,
+       		sample=sample[0],
+    	    docker=sentieon_docker,
+        	db_mills=db_mills,
+	        dbmills_dir=dbmills_dir,
+    	    dbsnp=dbsnp,
+        	dbsnp_dir=dbsnp_dir,
+	        tumor_recaled_bam=sample[1],
+    	    tumor_recaled_bam_index=sample[2],
+        	normal_recaled_bam=sample[3],
+	        normal_recaled_bam_index=sample[4],
+    	    disk_size=disk_size,
+        	cluster_config=cluster_config
+    	}
+
+    	call TNseq.TNseq as TNseq {
+            input:
+            SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+            SENTIEON_LICENSE=SENTIEON_LICENSE,
+            PONmode=PONmode,
+            fasta=fasta,
+            ref_dir=ref_dir,
+            corealigner_bam=corealigner.corealigner_bam,
+            corealigner_bam_index=corealigner.corealigner_bam_index,
+            dbsnp=dbsnp,
+            dbsnp_dir=dbsnp_dir,
+            tumor_name=sample[0] + "_tumor",
+            normal_name=sample[0] + "_normal",
+            cosmic_vcf=cosmic_vcf,
+            cosmic_dir=cosmic_dir,
+            panel_of_normal_vcf = bcftools.panel_of_normal_vcf,
+            docker=sentieon_docker,
+            sample=sample[0],
+            disk_size=disk_size,
+            cluster_config=cluster_config
+        }
+    
+        call TNscope.TNscope as TNscope {
+            input:
+            SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+            SENTIEON_LICENSE=SENTIEON_LICENSE,
+            PONmode=PONmode,
+            fasta=fasta,
+            ref_dir=ref_dir,
+            corealigner_bam=corealigner.corealigner_bam,
+            corealigner_bam_index=corealigner.corealigner_bam_index,
+            dbsnp=dbsnp,
+            dbsnp_dir=dbsnp_dir,
+            tumor_name=sample[0] + "_tumor",
+            normal_name=sample[0] + "_normal",
+            cosmic_vcf=cosmic_vcf,
+            cosmic_dir=cosmic_dir,
+            panel_of_normal_vcf = bcftools.panel_of_normal_vcf,
+            docker=sentieon_docker,
+            sample=sample[0],
+            disk_size=disk_size,
+            cluster_config=cluster_config
+        }
+
+        if (set_annovar){
+            call annovar.annovar as annovar {
+                input:
+                docker=annovar_docker,
+                database=database,
+                tnscope_vcf_file=TNscope.TNscope_vcf,
+                tnseq_vcf_file=TNseq.TNseq_vcf,
+                sample=sample[0],
+                cluster_config=cluster_config,
+                disk_size=disk_size
+            }
+        }
+    
+        if (set_vcf2maf){
+            call vcf2maf.vcf2maf as vcf2maf {
+                input:
+                docker=vcf2maf_r_docker,
+                multianno_tnscope_txt=annovar.multianno_tnscope_txt,
+                multianno_tnseq_txt=annovar.multianno_tnseq_txt,
+                sample=sample[0],
+                cluster_config=cluster_config,
+                disk_size=disk_size
+            }
+        }
+	}
+}