first commit

README.md

@@ -0,0 +1,2 @@
+WES capture efficiency (from bam file)
+

defaults

@@ -0,0 +1,9 @@
+{
+  "SENTIEON_INSTALL_DIR": "/opt/sentieon-genomics",
+  "fasta": "GRCh38.d1.vd1.fa",
+  "picard_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/picard:2.20.2",
+  "sentieon_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/sentieon-genomics:v2019.11.28",
+  "disk_size": "200",
+  "SMALLcluster_config": "OnDemand bcs.ps.g.xlarge img-ubuntu-vpc",
+  "ref_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/"
+}

inputs

@@ -0,0 +1,17 @@
+{
+  "{{ project_name }}.SENTIEON_INSTALL_DIR": "{{ SENTIEON_INSTALL_DIR }}",
+  "{{ project_name }}.Dedup_bam": "{{ Dedup_bam }}",
+  "{{ project_name }}.fasta": "{{ fasta }}",
+  "{{ project_name }}.disk_size": "{{ disk_size }}",
+  "{{ project_name }}.sentieon_docker": "sentieon_docker",
+  "{{ project_name }}.picard_docker": "picard_docker",
+  "{{ project_name }}.bed": "{{ bed }}",
+  "{{ project_name }}.SMALLcluster_config": "{{ SMALLcluster_config }}",
+  "{{ project_name }}.interval_list_name": "{{ interval_list_name }}",
+  "{{ project_name }}.Dedup_bam_index": "{{ Dedup_bam_index }}",
+  "{{ project_name }}.sample": "{{ sample }}",
+  "{{ project_name }}.ref_dir": "{{ ref_dir }}"
+}
+
+
+

tasks/bed_to_interval_list.wdl

@@ -0,0 +1,30 @@
+task bed_to_interval_list {
+
+	File reference_bed_dict
+    File bed
+	String interval_list_name
+	String docker
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e 
+       java -jar /usr/bin/picard/picard.jar BedToIntervalList \
+       I=${bed} \
+       O=${interval_list_name}.txt \
+       SD=${reference_bed_dict}
+	>>>
+
+	runtime {
+		docker:docker
+    	cluster: cluster_config
+    	systemDisk: "cloud_ssd 40"
+    	dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/" 
+	}
+
+	output {
+		File interval_list = "${interval_list_name}.txt"
+	}
+}

tasks/deduped_Metrics.wdl

@@ -0,0 +1,47 @@
+task deduped_Metrics {
+
+    File ref_dir
+    File bed
+	File interval_list
+	String SENTIEON_INSTALL_DIR
+	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=192.168.0.55:8990
+		nt=$(nproc)
+		${SENTIEON_INSTALL_DIR}/bin/sentieon driver -r ${ref_dir}/${fasta} -t $nt -i ${Dedup_bam} --interval ${bed} --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 --algo QualityYield ${sample}_deduped_QualityYield.txt --algo WgsMetricsAlgo ${sample}_deduped_WgsMetricsAlgo.txt --algo HsMetricAlgo --targets_list ${interval_list} --baits_list ${interval_list} ${sample}_deduped_HsMetricAlgo.txt
+	>>>
+
+	runtime {
+		docker:docker
+    	cluster: cluster_config
+    	systemDisk: "cloud_ssd 40"
+    	dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/" 
+	}
+
+	output {
+		File deduped_coverage_metrics_sample_summary = "${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"
+		File deduped_mean_quality = "${sample}_deduped_mq_metrics.txt"
+		File deduped_qd_metrics = "${sample}_deduped_qd_metrics.txt"
+		File deduped_gc_summary = "${sample}_deduped_gc_summary.txt"
+		File deduped_gc_metrics = "${sample}_deduped_gc_metrics.txt"
+		File dedeuped_aln_metrics = "${sample}_deduped_aln_metrics.txt"
+		File deduped_is_metrics = "${sample}_deduped_is_metrics.txt"
+		File deduped_QualityYield = "${sample}_deduped_QualityYield.txt"
+		File deduped_wgsmetrics = "${sample}_deduped_WgsMetricsAlgo.txt"
+		File deduped_HsMetricAlgo = "${sample}_deduped_HsMetricAlgo.txt"
+	}
+}

tasks/reference_dict.wdl

@@ -0,0 +1,28 @@
+task reference_dict {
+
+    File ref_dir
+	String fasta
+	String docker
+	String cluster_config
+	String disk_size
+
+
+	command <<<
+		set -o pipefail
+		set -e 
+        java -jar /usr/bin/picard/picard.jar CreateSequenceDictionary \ 
+        R=${ref_dir}/${fasta}\ 
+        O=reference_bed_dict.txt
+	>>>
+
+	runtime {
+		docker:docker
+    	cluster: cluster_config
+    	systemDisk: "cloud_ssd 40"
+    	dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/" 
+	}
+
+	output {
+		File reference_bed_dict= "reference_bed_dict.txt"  
+	}
+}

workflow.wdl

@@ -0,0 +1,54 @@
+import "./tasks/reference_dict.wdl" as reference_dict
+import "./tasks/bed_to_interval_list.wdl" as bed_to_interval_list
+import "./tasks/deduped_Metrics.wdl" as deduped_Metrics
+
+workflow   project_name  {
+    File ref_dir
+	String fasta
+	String SMALLcluster_config
+	String disk_size
+    String picard_docker
+    String sentieon_docker
+    File bed
+	String interval_list_name
+	String SENTIEON_INSTALL_DIR
+	String sample
+	File Dedup_bam
+	File Dedup_bam_index
+
+
+        call reference_dict.reference_dict as reference_dict {
+			input:
+			ref_dir=ref_dir,
+			fasta=fasta,
+			docker=picard_docker,
+			cluster_config=SMALLcluster_config,
+			disk_size=disk_size
+		}
+       	
+		call bed_to_interval_list.bed_to_interval_list as bed_to_interval_list {
+			input:
+			bed=bed,
+			interval_list_name=interval_list_name,
+			reference_bed_dict=reference_dict.reference_bed_dict,
+			docker=picard_docker,
+			cluster_config=SMALLcluster_config,
+			disk_size=disk_size
+		}
+
+		call deduped_Metrics.deduped_Metrics as deduped_Metrics {
+			input:
+			SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
+			fasta=fasta,
+			ref_dir=ref_dir,
+			bed=bed,
+			Dedup_bam=Dedup_bam,
+			Dedup_bam_index=Dedup_bam_index,
+			sample=sample,
+			docker=sentieon_docker,
+			interval_list=bed_to_interval_list.interval_list,
+			disk_size=disk_size,
+			cluster_config=SMALLcluster_config
+		}
+}
+