rtg

inputs

@@ -10,6 +10,7 @@
   "{{ project_name }}.SMALLcluster_config": "OnDemand bcs.ps.g.xlarge img-ubuntu-vpc",
   "{{ project_name }}.BIGcluster_config": "OnDemand bcs.a2.7xlarge img-ubuntu-vpc",
   "{{ project_name }}.LCL6": "{{ LCL6 }}",
+  "{{ project_name }}.sdf": "oss://pgx-reference-data/GRCh38.d1.vd1/GRCh38.d1.vd1.sdf/",
   "{{ project_name }}.MENDELIANdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1",
   "{{ project_name }}.LCL7": "{{ LCL7 }}",
   "{{ project_name }}.DIYdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/high_confidence_call_manuscript:v1.4",

tasks/benchmark.wdl

@@ -1,7 +1,7 @@
 task benchmark {
 	File vcf
 	File benchmarking_dir
-	File ref_dir
+	File sdf
 	String sample = basename(vcf,".raw.vcf.gz")
 	String fasta
 	String docker
@@ -14,29 +14,37 @@ task benchmark {
 		set -e
 		nt=$(nproc)
 		mkdir -p /cromwell_root/tmp
-		cp -r ${ref_dir} /cromwell_root/tmp/
 		cp -r ${benchmarking_dir} /cromwell_root/tmp/
-
-		export HGREF=/cromwell_root/tmp/reference_data/GRCh38.d1.vd1.fa
+        ls /cromwell_root/tmp/reference_datasets_v202103/ > files.txt
 
 		gunzip ${vcf} -c > unzip.vcf
  
 		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip unzip.vcf -c > ${sample}.rtg.vcf.gz
 		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.rtg.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${sample}.rtg.vcf.gz -o ${sample}.rtg.SNV.vcf.gz --snps-only
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${sample}.rtg.vcf.gz -o ${sample}.rtg.INDEL.vcf.gz --non-snps-only
+
 
 		if [[ ${sample} =~ "LCL5" ]];then
-			/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
 	    elif [[ ${sample} =~ "LCL6" ]]; then
-	    	/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
         elif [[ ${sample} =~ "LCL7" ]]; then
-        	/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
 	    elif [[ ${sample} =~ "LCL8" ]]; then
-			/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.vcf.gz ${sample}.rtg.vcf.gz -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_INDEL -t ${sdf} --bed-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
         else
         	echo "only for quartet samples"
         fi		
 
-        ls /cromwell_root/tmp/reference_datasets_v202103/ > files.txt
+
+        cat ${sample}_SNV/summary.txt ${sample}_SNV_precision_recall.txt
+        cat ${sample}_INDEL/summary.txt ${sample}_INDEL_precision_recall.txt
+        
 	>>>
 
 	runtime {
@@ -49,16 +57,8 @@ task benchmark {
 	output {
 		File rtg_vcf = "${sample}.rtg.vcf.gz"
 		File rtg_vcf_index = "${sample}.rtg.vcf.gz.tbi"
-		File gzip_vcf = "${sample}.vcf.gz"
-		File gzip_vcf_index = "${sample}.vcf.gz.tbi"
-		File roc_all_csv = "${sample}.roc.all.csv.gz"
-		File roc_indel = "${sample}.roc.Locations.INDEL.csv.gz"
-		File roc_indel_pass = "${sample}.roc.Locations.INDEL.PASS.csv.gz"
-		File roc_snp = "${sample}.roc.Locations.SNP.csv.gz"
-		File roc_snp_pass = "${sample}.roc.Locations.SNP.PASS.csv.gz"
-		File summary = "${sample}.summary.csv"
-		File extended = "${sample}.extended.csv"
-		File metrics = "${sample}.metrics.json.gz"
+		File SNV_result = "${sample}_SNV_precision_recall.txt"
+		File Indel_result = "${sample}_INDEL_precision_recall.txt"
 		File file_list = "files.txt"
 	}
 }

workflow.wdl

@@ -19,6 +19,7 @@ workflow {{ project_name }} {
 
 	String fasta
 	File ref_dir
+	File sdf
 
 	File benchmarking_dir
 	File rename
@@ -33,7 +34,7 @@ workflow {{ project_name }} {
 		input:
 		vcf=LCL5,
 		benchmarking_dir=benchmarking_dir,
-		ref_dir=ref_dir,
+		sdf=sdf,
 		fasta=fasta,
 		docker=BENCHMARKdocker,
 		cluster_config=BIGcluster_config,
@@ -44,7 +45,7 @@ workflow {{ project_name }} {
 		input:
 		vcf=LCL6,
 		benchmarking_dir=benchmarking_dir,
-		ref_dir=ref_dir,
+		sdf=sdf,
 		fasta=fasta,
 		docker=BENCHMARKdocker,
 		cluster_config=BIGcluster_config,
@@ -55,7 +56,7 @@ workflow {{ project_name }} {
 		input:
 		vcf=LCL7,
 		benchmarking_dir=benchmarking_dir,
-		ref_dir=ref_dir,
+		sdf=sdf,
 		fasta=fasta,
 		docker=BENCHMARKdocker,
 		cluster_config=BIGcluster_config,
@@ -66,7 +67,7 @@ workflow {{ project_name }} {
 		input:
 		vcf=LCL8,
 		benchmarking_dir=benchmarking_dir,
-		ref_dir=ref_dir,
+		sdf=sdf,
 		fasta=fasta,
 		docker=BENCHMARKdocker,
 		cluster_config=BIGcluster_config,