validation

inputs

@@ -1,5 +1,6 @@
 {
   "{{ project_name }}.satellite": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/Satellite.3.bed",
+  "{{ project_name }}.D5_10X": "oss://pgx-result/renluyao/manuscript_v3.0/10X/LCL5_10X.vcf.gz",
   "{{ project_name }}.mhc": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/mhc.hg38.bed",
   "{{ project_name }}.disk_size": "100",
   "{{ project_name }}.vcf_idx": "{{ vcf_idx }}",
@@ -8,10 +9,12 @@
   "{{ project_name }}.sample_name": "{{ sample_name }}",
   "{{ project_name }}.del_breakpoint": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/Tier1.del.breakpoint",
   "{{ project_name }}.ins_breakpoint": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/Tier1.ins.breakpoint",
+  "{{ project_name }}.sdf": "oss://pgx-reference-data/GRCh38.d1.vd1/GRCh38.d1.vd1.sdf/"
   "{{ project_name }}.LINE": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/LINE.3.bed",
   "{{ project_name }}.cluster_config": "OnDemand bcs.a2.large img-ubuntu-vpc",
   "{{ project_name }}.LTR": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/LTR.3.bed",
   "{{ project_name }}.simple_repeat": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/simple_repeat.3.bed",
+  "{{ project_name }}.D5_10X_idx": "oss://pgx-result/renluyao/manuscript_v3.0/10X/LCL5_10X.vcf.gz.tbi",
   "{{ project_name }}.vcf": "{{ vcf }}",
   "{{ project_name }}.confidence_bed": "oss://pgx-result/renluyao/manuscript_v3.0/reference_dataset_v202011/Quartet.callable.voted.collapse.bed",
   "{{ project_name }}.low_complexity": "oss://pgx-result/renluyao/manuscript_v3.0/difficult_region/Low_complexity.3.bed",

tasks/validation.wdl

@@ -0,0 +1,111 @@
+task validation {
+	File sub_vcf
+	File sub_vcf_idx
+
+	File satellite
+	File simple_repeat
+	File SINE
+	File LINE
+	File LTR
+	File low_complexity
+	File SD
+	File ins_breakpoint
+	File del_breakpoint
+	File mhc
+
+	File sdf
+	File D5_10X
+	File D5_10X_idx
+
+	String sample_name
+	String tag
+	String docker
+	String cluster_config
+	String disk_size
+	
+	command <<<
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.satellite.vcf.gz --include-bed=${satellite}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.simple_repeat.vcf.gz --include-bed=${simple_repeat}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.SINE.vcf.gz --include-bed=${SINE}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.LINE.vcf.gz --include-bed=${LINE}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.LTR.vcf.gz --include-bed=${LTR}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.low_complexity.vcf.gz --include-bed=${low_complexity}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.SD.vcf.gz --include-bed=${SD}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.ins_breakpoint.vcf.gz --include-bed=${ins_breakpoint}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.del_breakpoint.vcf.gz --include-bed=${del_breakpoint}
+
+		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.mhc.vcf.gz --include-bed=${mhc}
+
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.satellite.vcf.gz -o ${sample_name}.${tag}.satellite -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.simple_repeat.vcf.gz -o ${sample_name}.${tag}.simple_repeat -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.SINE.vcf.gz -o ${sample_name}.${tag}.SINE -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.LINE.vcf.gz -o ${sample_name}.${tag}.LINE -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.LTR.vcf.gz -o ${sample_name}.${tag}.LTR -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.low_complexity.vcf.gz -o ${sample_name}.${tag}.low_complexity -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.SD.vcf.gz -o ${sample_name}.${tag}.SD -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.ins_breakpoint.vcf.gz -o ${sample_name}.${tag}.ins_breakpoint -t ${sdf}
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.del_breakpoint.vcf.gz -o ${sample_name}.${tag}.del_breakpoint -t ${sdf}	
+
+		rtg vcfeval -b ${D5_10X} -c ${sample_name}.${tag}.mhc.vcf.gz -o ${sample_name}.${tag}.mhc -t ${sdf}	
+
+
+		cp ${sample_name}.${tag}.satellite/summary.txt > ${sample_name}.${tag}.satellite.summary.txt
+		cp ${sample_name}.${tag}.simple_repeat/summary.txt > ${sample_name}.${tag}.simple_repeat.summary.txt
+		cp ${sample_name}.${tag}.SINE/summary.txt > ${sample_name}.${tag}.SINE.summary.txt
+		cp ${sample_name}.${tag}.LINE/summary.txt > ${sample_name}.${tag}.LINE.summary.txt
+		cp ${sample_name}.${tag}.LTR/summary.txt > ${sample_name}.${tag}.LTR.summary.txt
+		cp ${sample_name}.${tag}.low_complexity/summary.txt > ${sample_name}.${tag}.low_complexity.summary.txt
+		cp ${sample_name}.${tag}.SD/summary.txt > ${sample_name}.${tag}.SD.summary.txt
+		cp ${sample_name}.${tag}.ins_breakpoint/summary.txt > ${sample_name}.${tag}.ins_breakpoint.summary.txt
+		cp ${sample_name}.${tag}.del_breakpoint/summary.txt > ${sample_name}.${tag}.del_breakpoint.summary.txt
+		cp ${sample_name}.${tag}.mhc/summary.txt > ${sample_name}.${tag}.mhc.summary.txt
+
+
+	>>>
+
+	runtime {
+		docker:docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File satellite_vcf = "${sample_name}.${tag}.satellite.vcf.gz"
+		File simple_repeat_vcf = "${sample_name}.${tag}.simple_repeat.vcf.gz"
+		File SINE_vcf = "${sample_name}.${tag}.SINE.vcf.gz"
+		File LINE_vcf = "${sample_name}.${tag}.LINE.vcf.gz"
+		File LTR_vcf = "${sample_name}.${tag}.LTR.vcf.gz"
+		File low_complexity_vcf = "${sample_name}.${tag}.low_complexity.vcf.gz"
+		File SD_vcf = "${sample_name}.${tag}.SD.vcf.gz"
+		File ins_breakpoint_vcf = "${sample_name}.${tag}.ins_breakpoint.vcf.gz"
+		File del_breakpoint_vcf = "${sample_name}.${tag}.del_breakpoint.vcf.gz"
+		File mhc_vcf = "${sample_name}.${tag}.mhc.vcf.gz"
+		File satellite_summary = "${sample_name}.${tag}.satellite.summary.txt"
+		File simple_repeat_summary = "${sample_name}.${tag}.simple_repeat.summary.txt"
+		File SINE_summary = "${sample_name}.${tag}.SINE.summary.txt"
+		File LINE_summary = "${sample_name}.${tag}.LINE.summary.txt"
+		File LTR_summary = "${sample_name}.${tag}.LTR.summary.txt"
+		File low_complexity_summary = "${sample_name}.${tag}.low_complexity.summary.txt"
+		File SD_summary = "${sample_name}.${tag}.SD.summary.txt"
+		File ins_breakpoint_summary = "${sample_name}.${tag}.ins_breakpoint.summary.txt"
+		File del_breakpoint_summary = "${sample_name}.${tag}.del_breakpoint.summary.txt"
+		File mhc_summary = "${sample_name}.${tag}.mhc.summary.txt"
+	}
+}

workflow.wdl

@@ -1,5 +1,5 @@
 import "./tasks/in_out.wdl" as in_out
-import "./tasks/region.wdl" as region
+import "./tasks/validation.wdl" as validation
 
 
 workflow {{ project_name }} {
@@ -20,6 +20,10 @@ workflow {{ project_name }} {
 	File del_breakpoint
 	File mhc
 
+	File sdf
+	File D5_10X
+	File D5_10X_idx
+
 	String sample_name
 
 	String docker
@@ -37,7 +41,7 @@ workflow {{ project_name }} {
 		disk_size=disk_size
 	}
 
-	call region.region as in_region {
+	call validation.validation as in_validation {
 		input:
 		sub_vcf=in_out.in_vcf,
 		sub_vcf_idx=in_out.in_vcf_idx,
@@ -51,6 +55,9 @@ workflow {{ project_name }} {
 		ins_breakpoint=ins_breakpoint,
 		del_breakpoint=del_breakpoint,
 		mhc=mhc,
+		sdf=sdf,
+		D5_10X=D5_10X,
+		D5_10X_idx=D5_10X_idx,
 		sample_name=sample_name,
 		tag="inside",
 		docker=docker,
@@ -58,7 +65,7 @@ workflow {{ project_name }} {
 		disk_size=disk_size
 	}
 
-	call region.region as out_region {
+	call validation.validation as out_validation {
 		input:
 		sub_vcf=in_out.out_vcf,
 		sub_vcf_idx=in_out.out_vcf_idx,
@@ -72,6 +79,9 @@ workflow {{ project_name }} {
 		ins_breakpoint=ins_breakpoint,
 		del_breakpoint=del_breakpoint,
 		mhc=mhc,
+		sdf=sdf,
+		D5_10X=D5_10X,
+		D5_10X_idx=D5_10X_idx,
 		sample_name=sample_name,
 		tag="outside",
 		docker=docker,