first commit

inputs

@@ -0,0 +1,11 @@
+{
+  "{{ project_name }}.D6_trio_raw_vcf": "{{ D6_trio_raw_vcf }}",
+  "{{ project_name }}.high_confidence_bed": "{{ high_confidence_bed }}",
+  "{{ project_name }}.family_name": "{{ family_name }}",
+  "{{ project_name }}.D5_trio_raw_vcf": "{{ D5_trio_raw_vcf }}",
+  "{{ project_name }}.BEDdocker": "registry-internal.cn-shanghai.aliyuncs.com/pgx-docker-registry/bedtools:v2.27.1",
+  "{{ project_name }}.disk_size": "200",
+  "{{ project_name }}.SMALLcluster_config": "OnDemand bcs.ps.g.xlarge img-ubuntu-vpc",
+  "{{ project_name }}.family_raw_vcf": "{{ family_raw_vcf }}",
+  "{{ project_name }}.DIYdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/high_confidence_call_manuscript:v1.4"
+}

tasks/merge_mendelian.wdl

@@ -0,0 +1,32 @@
+task merge_mendelian {
+	File D5_in
+	File D6_in
+	File family_in
+	File D5_out
+	File D6_out
+	File family_out
+	String family_name
+	String docker
+	String cluster_config
+	String disk_size
+	
+	command <<<
+		python /opt/merge_two_family_with_genotype.py -LCL5 ${D5_in} -LCL6 ${D6_in} -genotype ${family_in} -family ${family_name}.inside
+
+		python /opt/merge_two_family_with_genotype.py -LCL5 ${D5_out} -LCL6 ${D6_out} -genotype ${family_out} -family ${family_name}.outside
+
+	>>>
+
+	runtime {
+		docker:docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File mendelian_in = "${family_name}.inside.txt"
+		File mendelian_in_summary = "${family_name}.inside.summary.txt"
+		File mendelian_out = "${family_name}.outside.txt"
+		File mendelian_out_summary = "${family_name}.outside.summary.txt"
+	}
+}

tasks/split_file.wdl

@@ -0,0 +1,68 @@
+task split_file {
+	File D5_trio_raw_vcf
+	File D6_trio_raw_vcf
+	File family_raw_vcf
+	File high_confidence_bed
+	String family_name
+	String docker
+	String cluster_config
+	String disk_size
+	
+	command <<<
+		cat ${D5_trio_raw_vcf} | grep '#CHROM' > D5.header
+		cat ${D6_trio_raw_vcf} | grep '#CHROM' > D6.header
+		cat ${family_raw_vcf} | grep '#CHROM' > family.header
+
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${D5_trio_raw_vcf} -b ${high_confidence_bed} > D5.in
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${D6_trio_raw_vcf} -b ${high_confidence_bed} > D6.in
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${family_raw_vcf} -b ${high_confidence_bed} > family.in
+
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${D5_trio_raw_vcf} -b ${high_confidence_bed} -v > D5.out
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${D6_trio_raw_vcf} -b ${high_confidence_bed} -v > D6.out
+		/opt/ccdg/bedtools-2.27.1/bin/bedtools intersect -a ${family_raw_vcf} -b ${high_confidence_bed} -v > family.out
+
+		cat D5.header D5.in > ${family_name}.D5.in.txt
+		cat D6.header D6.in > ${family_name}.D6.in.txt
+		cat family.header family.in > ${family_name}.family.in.txt
+
+		cat D5.header D5.out > ${family_name}.D5.out.txt
+		cat D6.header D6.out > ${family_name}.D6.out.txt
+		cat family.header family.out > ${family_name}.family.out.txt
+
+		cat ${family_name}.family.in.txt | awk '
+		    BEGIN { OFS = "\t" }
+		    NF > 2 && FNR > 1 { 
+		        for ( i=9; i<=NF; i++ ) { 
+		            split($i,a,":") ;$i = a[1];
+		        } 
+		    } 
+		    { print }
+		' > ${family_name}.family.in.gt.txt
+
+		cat ${family_name}.family.out.txt | awk '
+		    BEGIN { OFS = "\t" }
+		    NF > 2 && FNR > 1 { 
+		        for ( i=9; i<=NF; i++ ) { 
+		            split($i,a,":") ;$i = a[1];
+		        } 
+		    } 
+		    { print }
+		' > ${family_name}.family.out.gt.txt
+
+	>>>
+
+	runtime {
+		docker:docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File D5_in = "${family_name}.D5.in.txt"
+		File D6_in = "${family_name}.D6.in.txt"
+		File family_in = "${family_name}.family.in.gt.txt"
+		File D5_out = "${family_name}.D5.out.txt"
+		File D6_out = "${family_name}.D6.out.txt"
+		File family_out = "${family_name}.family.out.gt.txt"
+	}
+}

workflow.wdl

@@ -0,0 +1,45 @@
+import "./tasks/split_file.wdl" as split_file
+import "./tasks/merge_mendelian.wdl" as merge_mendelian
+
+workflow {{ project_name }} {
+
+	File D5_trio_raw_vcf
+	File D6_trio_raw_vcf
+	File family_raw_vcf
+	File high_confidence_bed
+
+	String BEDdocker
+	String DIYdocker
+	String family_name
+
+	String disk_size
+	String SMALLcluster_config
+
+	call split_file.split_file as split_file {
+		input:
+		D5_trio_raw_vcf=D5_trio_raw_vcf,
+		D6_trio_raw_vcf=D6_trio_raw_vcf,
+		family_raw_vcf=family_raw_vcf,
+		high_confidence_bed=high_confidence_bed,
+		family_name=family_name,
+		docker=BEDdocker,
+		cluster_config=SMALLcluster_config,
+		disk_size=disk_size
+	}
+
+	call merge_mendelian.merge_mendelian as merge_mendelian {
+		input:
+		D5_in=split_file.D5_in,
+		D6_in=split_file.D6_in,
+		family_in=split_file.family_in,
+		D5_out=split_file.D5_out,
+		D6_out=split_file.D6_out,
+		family_out=split_file.family_out,
+		family_name=family_name,
+		docker=DIYdocker,
+		cluster_config=SMALLcluster_config,
+		disk_size=disk_size
+	}
+}
+
+