first commit

defaults

@@ -0,0 +1,17 @@
+{
+    "fasta": "GRCh38.d1.vd1.fa",
+    "ref_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/",
+    "dbsnp": "dbsnp_146.hg38.vcf",
+    "dbsnp_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/",
+    "SENTIEON_INSTALL_DIR": "/opt/sentieon-genomics",
+    "SENTIEON_LICENSE": "192.168.0.55:8990",
+    "dbmills_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/",
+    "db_mills": "Mills_and_1000G_gold_standard.indels.hg38.vcf",
+    "sentieon_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/sentieon-genomics:v2019.11.28",
+    "database": "oss://pgx-reference-data/annovar_hg38/",
+    "regions": "oss://pgx-reference-data/reference/wes_bedfiles/agilent_v6/SureSelect_Human_All_Exon_V6_r2.bed",
+    "cosmic_vcf": "CosmicCodingMuts.hg38.v91.vcf",
+    "cosmic_dir": "oss://pgx-reference-data/reference/cosmic/",
+    "disk_size": "200",
+    "cluster_config": "OnDemand bcs.a2.3xlarge img-ubuntu-vpc"
+}

inputs

@@ -0,0 +1,20 @@
+{
+  "{{ project_name }}.sample_id": "{{ sample_id }}",
+  "{{ project_name }}.fastq_1": "{{ fastq_1 }}",
+  "{{ project_name }}.fastq_2": "{{ fastq_2 }}",
+  "{{ 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 }}.database": "{{ database }}",
+  "{{ project_name }}.regions": "{{ regions }}",
+  "{{ project_name }}.cosmic_vcf": "{{ cosmic_vcf }}",
+  "{{ project_name }}.cosmic_dir": "{{ cosmic_dir }}",
+  "{{ project_name }}.disk_size": "{{ disk_size }}",
+  "{{ project_name }}.cluster_config": "{{ cluster_config }}"
+}

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 +1,41 @@
+task PON {
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sample
+
+	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)
+		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} ${sample}_pon.vcf
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"	
+	}
+
+	output {
+		File pon_vcf = "${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 +1,62 @@
+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
+	File tumor_recaled_bam
+	File tumor_recaled_bam_index
+	String dbsnp
+	String disk_size
+
+	Boolean set_pon
+	String? cosmic_vcf
+	File? cosmic_dir
+	File? pon_vcf
+	
+	String pon_command = if set_pon then "--pon /cromwell_root/tmp/PON/$(basename ${pon_vcf}) --cosmic /cromwell_root/tmp/PON/${cosmic_vcf}" else ""
+	Boolean TN = if defined(corealigner_bam) then true else false
+	
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+
+		mkdir -p /cromwell_root/tmp/PON/
+		cp ${cosmic_dir}/${cosmic_vcf} /cromwell_root/tmp/PON/
+		cp ${pon_vcf} /cromwell_root/tmp/PON/
+		${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/PON/${cosmic_vcf}
+		${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/PON/$(basename ${pon_vcf})
+		
+		if ${TN}; then
+			${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} ${pon_command} ${sample}.TNscope.TN.vcf
+		else
+			${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${tumor_recaled_bam} --algo TNscope --tumor_sample ${tumor_name} --dbsnp ${dbsnp_dir}/${dbsnp} ${pon_command} ${sample}.TNscope.TN.vcf
+		fi
+		
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File TNscope_vcf= "${sample}.TNscope.TN.vcf"
+		File TNscope_vcf_index = "${sample}.TNscope.TN.vcf.idx"
+	}
+
+}

tasks/TNseq.wdl

@@ -0,0 +1,61 @@
+task TNseq {
+	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
+	File tumor_recaled_bam
+	File tumor_recaled_bam_index
+	String dbsnp
+	String disk_size
+
+	Boolean set_pon
+	String? cosmic_vcf
+	File? cosmic_dir
+	File? pon_vcf
+	
+	String pon_command = if set_pon then "--pon /cromwell_root/tmp/PON/$(basename ${pon_vcf}) --cosmic /cromwell_root/tmp/PON/${cosmic_vcf}" else ""
+	Boolean TN = if defined(corealigner_bam) then true else false
+
+	command <<<
+		set -o pipefail
+		set -e
+		export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+		nt=$(nproc)
+
+		mkdir -p /cromwell_root/tmp/PON/
+		cp ${cosmic_dir}/${cosmic_vcf} /cromwell_root/tmp/PON/
+		cp ${pon_vcf} /cromwell_root/tmp/PON/
+		${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/PON/${cosmic_vcf}
+		${SENTIEON_INSTALL_DIR}/bin/sentieon util vcfindex /cromwell_root/tmp/PON/$(basename ${pon_vcf})
+
+		if ${TN}; then
+			${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${corealigner_bam} --algo TNhaplotyper --tumor_sample ${tumor_name} --normal_sample ${normal_name} --dbsnp ${dbsnp_dir}/${dbsnp} ${pon_command} ${sample}.TNseq.TN.vcf
+		else
+			${SENTIEON_INSTALL_DIR}/bin/sentieon driver -t $nt -r ${ref_dir}/${fasta} -i ${tumor_recaled_bam} --algo TNhaplotyper --tumor_sample ${tumor_name} --dbsnp ${dbsnp_dir}/${dbsnp} ${pon_command} ${sample}.TNseq.TN.vcf
+		fi
+	>>>
+
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File TNseq_vcf= "${sample}.TNseq.TN.vcf"
+		File TNseq_vcf_index = "${sample}.TNseq.TN.vcf.idx"
+	}
+
+}
+
+

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 +1,57 @@
+import "./tasks/mapping.wdl" as mapping
+import "./tasks/Dedup.wdl" as Dedup
+
+
+workflow {{ project_name }} {
+	
+	String sample_id
+	File fastq_1
+	File fastq_2
+
+	String SENTIEON_INSTALL_DIR
+	String SENTIEON_LICENSE
+	String sentieon_docker
+	
+	File ref_dir
+	String fasta
+	File dbmills_dir
+	String db_mills
+	File dbsnp_dir
+	String dbsnp
+	File regions
+	String disk_size
+	String cluster_config
+	
+	File? cosmic_dir
+	String? cosmic_vcf
+	File? database
+
+
+	call mapping.mapping as mapping {
+		input: 
+		group=sample_id + '_' + type,
+		sample=sample_id + '_' + type,
+		fastq_1=fastq_1,
+		fastq_2=fastq_2,
+		SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+		SENTIEON_LICENSE=SENTIEON_LICENSE,
+		pl="ILLUMINAL",
+		fasta=fasta,
+		ref_dir=ref_dir,
+		docker=sentieon_docker,
+		disk_size=disk_size,
+		cluster_config=cluster_config
+	}
+
+	call Dedup.Dedup as Dedup {
+		input:
+		SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+		SENTIEON_LICENSE=SENTIEON_LICENSE,
+		sorted_bam=mapping.sorted_bam,
+		sorted_bam_index=mapping.sorted_bam_index,
+		sample=sample_id + '_' + type,
+		docker=sentieon_docker,
+		disk_size=disk_size,
+		cluster_config=cluster_config
+	}
+}