split fastq and vcf

tasks/benchmark.wdl

@@ -19,17 +19,41 @@ task benchmark {
 
 		export HGREF=/cromwell_root/tmp/reference_data/GRCh38.d1.vd1.fa
 
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${filtered_vcf} -c > ${sample}.rtg.vcf.gz
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.rtg.vcf.gz
+
+		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL5" > LCL5_name
+		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL6" > LCL6_name
+		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL7" > LCL7_name
+		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL8" > LCL8_name
+
 
 		if [[ ${sample} =~ "LCL5" ]];then
-			/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.vcf ${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/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.vcf ${filtered_vcf} -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			cat ${filtered_vcf} | grep '##' > header
+			cat ${filtered_vcf} | grep -v '#' > body
+			cat header LCL5_name body > LCL5.vcf
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL5.vcf -c > ${sample}.reformed.vcf.gz
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.reformed.vcf.gz
 		elif [[ ${sample} =~ "LCL6" ]]; then
-		    /opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.vcf ${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/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.vcf ${filtered_vcf} -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			cat ${filtered_vcf} | grep '##' > header
+			cat ${filtered_vcf} | grep -v '#' > body
+			cat header LCL6_name body > LCL6.vcf
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL6.vcf -c > ${sample}.reformed.vcf.gz
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.reformed.vcf.gz
 	    elif [[ ${sample} =~ "LCL7" ]]; then
-	        /opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.vcf ${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/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.vcf ${filtered_vcf} -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			cat ${filtered_vcf} | grep '##' > header
+			cat ${filtered_vcf} | grep -v '#' > body
+			cat header LCL7_name body > LCL7.vcf
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL7.vcf -c > ${sample}.reformed.vcf.gz
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.reformed.vcf.gz
 		elif [[ ${sample} =~ "LCL8" ]]; then
-			/opt/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.vcf ${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/hap.py/bin/hap.py /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.vcf ${filtered_vcf} -f /cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed --threads $nt -o ${sample} -r ${ref_dir}/${fasta}
+			cat ${filtered_vcf} | grep '##' > header
+			cat ${filtered_vcf} | grep -v '#' > body
+			cat header LCL8_name body > LCL8.vcf
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL8.vcf -c > ${sample}.reformed.vcf.gz
+			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.reformed.vcf.gz
 	    else
 	        echo "only for quartet samples"
 	    fi
@@ -43,8 +67,8 @@ task benchmark {
 	}
 
 	output {
-		File rtg_vcf = "${sample}.rtg.vcf.gz"
-		File rtg_vcf_index = "${sample}.rtg.vcf.gz.tbi"
+		File rtg_vcf = "${sample}.reformed.vcf.gz"
+		File rtg_vcf_index = "${sample}.reformed.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"

tasks/mendelian.wdl

@@ -11,14 +11,14 @@ task mendelian {
 		export LD_LIBRARY_PATH=/opt/htslib-1.9
 		nt=$(nproc)
 
-		echo -e "${family_name}\tM8\t0\t0\t2\t-9\n${family_name}\tF7\t0\t0\t1\t-9\n${family_name}\tD5\tF7\tM8\t2\t-9" > ${family_name}.D5.ped
+		echo -e "${family_name}\tLCL8\t0\t0\t2\t-9\n${family_name}\tLCL7\t0\t0\t1\t-9\n${family_name}\tLCL5\tLCL7\tLCL8\t2\t-9" > ${family_name}.D5.ped
 
 		mkdir VBT_D5
 		/opt/VBT-TrioAnalysis/vbt mendelian -ref ${ref_dir}/${fasta} -mother ${family_vcf} -father ${family_vcf} -child ${family_vcf} -pedigree ${family_name}.D5.ped -outDir VBT_D5 -out-prefix ${family_name}.D5 --output-violation-regions -thread-count $nt
 
 		cat VBT_D5/${family_name}.D5_trio.vcf > ${family_name}.D5.vcf
 
-		echo -e "${family_name}\tM8\t0\t0\t2\t-9\n${family_name}\tF7\t0\t0\t1\t-9\n${family_name}\tD6\tF7\tM8\t2\t-9" > ${family_name}.D6.ped
+		echo -e "${family_name}\tLCL8\t0\t0\t2\t-9\n${family_name}\tLCL7\t0\t0\t1\t-9\n${family_name}\tLCL6\tLCL7\tLCL8\t2\t-9" > ${family_name}.D6.ped
 
 		mkdir VBT_D6
 		/opt/VBT-TrioAnalysis/vbt mendelian -ref ${ref_dir}/${fasta} -mother ${family_vcf} -father ${family_vcf} -child ${family_vcf} -pedigree ${family_name}.D6.ped -outDir VBT_D6 -out-prefix ${family_name}.D6 --output-violation-regions -thread-count $nt