jaccard index

inputs

@@ -4,23 +4,31 @@
   "{{ project_name }}.fasta": "GRCh38.d1.vd1.fa",
   "{{ project_name }}.fastqc.disk_size": "150",
   "{{ project_name }}.benchmark.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
+  "{{ project_name }}.mergeJI.disk_size": "100",
   "{{ project_name }}.fastqscreen.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
   "{{ project_name }}.fastqc.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
+  "{{ project_name }}.inputJIpiarsFile": "{{ inputJIpiarsFile }}",
   "{{ project_name }}.benchmark.disk_size": "150",
   "{{ project_name }}.vcfstat.disk_size": "100",
   "{{ project_name }}.fastqc.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/fastqc:v0.11.5",
+  "{{ project_name }}.mergeJI.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-tools:latest",
   "{{ project_name }}.benchmark.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-hap:latest",
   "{{ project_name }}.inputSamplesFile": "{{ inputSamplesFile }}",
+  "{{ project_name }}.mergeJI.cluster_config": "OnDemand ecs.sn1ne.xlarge img-ubuntu-vpc",
   "{{ project_name }}.fastqscreen.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/fastqscreen:0.12.0",
   "{{ project_name }}.mergeNum.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/gatk:v2019.01",
+  "{{ project_name }}.JI.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-hap:latest",
   "{{ project_name }}.screen_ref_dir": "oss://pgx-reference-data/fastq_screen_reference/",
   "{{ project_name }}.mergeNum.disk_size": "100",
+  "{{ project_name }}.JI.disk_size": "100",
   "{{ project_name }}.fastq_screen_conf": "oss://pgx-reference-data/fastq_screen_reference/fastq_screen.conf",
   "{{ project_name }}.multiqc.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
+  "{{ project_name }}.sdf": "oss://chinese-quartet/quartet-storage-data/reference_data/GRCh38.d1.vd1.sdf/",
   "{{ project_name }}.multiqc.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/multiqc:v1.8",
   "{{ project_name }}.mergeNum.cluster_config": "OnDemand ecs.sn1ne.xlarge img-ubuntu-vpc",
   "{{ project_name }}.bamqc.cluster_config": "OnDemand ecs.sn1ne.8xlarge img-ubuntu-vpc",
   "{{ project_name }}.vcfstat.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
+  "{{ project_name }}.JI.cluster_config": "OnDemand ecs.sn1ne.4xlarge img-ubuntu-vpc",
   "{{ project_name }}.fastqscreen.disk_size": "100",
   "{{ project_name }}.bamqc.disk_size": "500",
   "{{ project_name }}.multiqc.disk_size": "100",

tasks/jaccard_index.wdl

@@ -3,18 +3,22 @@ task JI {
 	File vcf_b
 	File dir_name
 	File sdf
+	String name_a = basename(vcf_a,".vcf")
+	String name_b = basename(vcf_b,".vcf")
 	String docker
 	String cluster_config
 	String disk_size
 	
 	command <<<
-		touch number
-		cat ${filelist}| while read a b c d
-		do
-			rtg vcfeval -b $a -c $b -o $c -t ${sdf}
-			cat $c/summary.txt | sed -n '4,4p' | sed 's/\s\+/\t/g'| cut -f4-6 >> number
-		done
-		paste ${filelist} number > result.txt
+		set -o pipefail
+		set -e
+
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${vcf_a} -c > ${name_a}.rtg.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${vcf_b}.rtg.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${vcf_a} -c > ${name_b}.rtg.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${vcf_b}.rtg.vcf.gz		
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b ${vcf_a} -c ${vcf_b} -o ${dir_name} -t ${sdf}
+
 	>>>
 
 	runtime {
@@ -24,6 +28,6 @@ task JI {
 		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
 	}
 	output {
-		File summary = "result.txt"
+		File JI_summary = "${dir_name}/summary.txt"
 	}
 }

tasks/mergeJI.wdl

@@ -0,0 +1,26 @@
+task mergeJI {
+	Array[File] JI_summary
+	File inputJIpiarsFile
+	String docker
+	String cluster_config
+	String disk_size
+	
+	command <<<
+		touch number
+		for i in ${sep=" " JI_summary}
+		do
+			cat $i | sed -n '4,4p' | sed 's/\s\+/\t/g'| cut -f4-6 >> number
+		done
+		paste ${inputJIpiarsFile} number > result.txt
+	>>>
+
+	runtime {
+		docker:docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File JI_all = "result.txt"
+	}
+}

workflow.wdl

@@ -5,20 +5,22 @@ import "./tasks/benchmark.wdl" as benchmark
 import "./tasks/multiqc.wdl" as multiqc
 import "./tasks/vcfstat.wdl" as vcfstat
 import "./tasks/mergeNum.wdl" as mergeNum
-#import "./tasks/jaccard_index.wdl" as JI
+import "./tasks/jaccard_index.wdl" as JI
+import "./tasks/mergeJI.wdl" as mergeJI
 
 
 workflow {{ project_name }} {
 
 	File inputSamplesFile
 	Array[Array[File]] inputSamples = read_tsv(inputSamplesFile)
-#	File inputJIpiarsFile
-#	Array[Array[File]] inputJIpairs = read_tsv(inputJIpiarsFile)
+	File inputJIpiarsFile
+	Array[Array[File]] inputJIpairs = read_tsv(inputJIpiarsFile)
 	File screen_ref_dir
 	File fastq_screen_conf
 	File benchmarking_dir
 	File ref_dir
 	String fasta
+	File sdf
 
 	scatter (sample in inputSamples){
 		call fastqc.fastqc as fastqc {
@@ -54,8 +56,7 @@ workflow {{ project_name }} {
 			input:
 			rtg_vcf=benchmark.rtg_vcf,
 			rtg_vcf_index=benchmark.rtg_vcf_index
-		}
-
+		}	
 
 	}
 	call multiqc.multiqc as multiqc {
@@ -73,5 +74,21 @@ workflow {{ project_name }} {
 		vcfnumber=vcfstat.vcfnumber
 	}
 
+	scatter (pair in inputJIpairs) {
+		call JI.JI as JI {
+			input:
+			vcf_a=pair[0],
+			vcf_b=pair[1],
+			dir_name=pair[2],
+			sdf=sdf
+		}	
+	}
+
+	call mergeJI.mergeJI as mergeJI {
+		input:
+		JI_summary=JI.JI_summary,
+		inputJIpiarsFile=inputJIpiarsFile
+	}
+
 }