sister and family seprate

README.md

@@ -23,7 +23,7 @@ choppy install renluyao/high_confidence_calls_intergration
 
 Quartet家系的四个样本的DNA标准物质被送往多加测序公司，用PCRfree或者PCR的建库方法建立3个技术重复，用多个测序平台（Illumina XTen、Illumina Novaseq、BGISEQ-500、BGISEQ-2000和BGISEQ-T7）进行测序。获得的原始数据用多个分析流程进行分析（比对软件：BWA-mem、Novoalign、Bowtie2，突变检出软件：HaplotyperCaller、Strelka2、FreeBayes）。
 
-进过以上各种因素的组合每个样本得到216个VCF文件，本APP的目的是整合这些VCF文件的结果得到一组可信度较高的突变。
+经过以上各种因素的组合每个样本得到216个VCF文件，本APP的目的是整合这些VCF文件的结果得到一组可信度较高的突变。
 
 ## 流程与参数
 

inputs

@@ -1,7 +1,8 @@
 {
   "{{ project_name }}.fasta": "GRCh38.d1.vd1.fa",
+  "{{ project_name }}.sister.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1",
   "{{ project_name }}.mendelian.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/vbt:v1.1",
-  "{{ project_name }}.disk_size": "100",
+  "{{ project_name }}.disk_size": "150",
   "{{ project_name }}.merge.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-tools:latest",
   "{{ project_name }}.zipIndex.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-tools:latest",
   "{{ project_name }}.inputSamplesFile": "{{ inputSamplesFile }}",

tasks/mendelian.wdl

@@ -1,10 +1,7 @@
 task mendelian {
-	File LCL5_vcf
-	File LCL6_vcf
+	File sister_vcf
 	File LCL7_vcf
 	File LCL8_vcf
-	String LCL5_name
-	String LCL6_name
 	String LCL7_name
 	String LCL8_name
 	String family_name
@@ -17,18 +14,9 @@ task mendelian {
 	command <<<
 		export LD_LIBRARY_PATH=/opt/htslib-1.9
 
-		mkdir sister
-
-		/opt/VBT-TrioAnalysis/vbt varcomp -called ${LCL5_vcf} -base ${LCL6_vcf} -ref ${ref_dir}/${fasta} -outDir sister -filter none
-
-		mv sister/TPBase.vcf ${family_name}.sister.consistent.vcf
-		mv sister/FP.vcf ${family_name}.LCL5.uniq.vcf
-		mv sister/FN.vcf ${family_name}.LCL6.uniq.vcf
-		mv sister/log.txt ${family_name}.sister.vbt.log.txt
-
 		mkdir VBT
 
-		/opt/VBT-TrioAnalysis/vbt mendelian -ref ${ref_dir}/${fasta} -mother ${LCL8_vcf} -father ${LCL7_vcf} -child ${family_name}.sister.consistent.vcf -outDir VBT -out-prefix ${family_name} --output-violation-regions
+		/opt/VBT-TrioAnalysis/vbt mendelian -ref ${ref_dir}/${fasta} -mother ${LCL8_vcf} -father ${LCL7_vcf} -child ${sister_vcf} -outDir VBT -out-prefix ${family_name} --output-violation-regions
 
 		cat VBT/${family_name}_trio.vcf | grep '#' | cut -f1-9,10 > mother_header
 		cat VBT/${family_name}_trio.vcf | grep -v '#' | cut -f1-9,10 | grep 'MD=1' | grep -v '0/0' | cat mother_header - > ${LCL8_name}.sister.mendelian.gt.vcf
@@ -48,10 +36,6 @@ task mendelian {
 		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
 	}
 	output {
-		File sister_consistent_vcf = "${family_name}.sister.consistent.vcf"
-		File LCL5_uniq_vcf = "${family_name}.LCL5.uniq.vcf"
-		File LCL6_uniq_vcf = "${family_name}.LCL6.uniq.vcf"
-		File sister_log = "${family_name}.sister.vbt.log.txt"
 		Array[File] vbt_mendelian = glob("VBT/*")
 		File mother_vcf = "${LCL8_name}.sister.mendelian.gt.vcf"
 		File father_vcf = "${LCL7_name}.sister.mendelian.gt.vcf"

tasks/sister.wdl

@@ -9,6 +9,8 @@ task sister {
 	String disk_size
 	
 	command <<<
+		export LD_LIBRARY_PATH=/opt/htslib-1.9
+
 		mkdir sister
 		vbt varcomp -called ${LCL5_vcf} -base ${LCL6_vcf} -ref ${ref_dir}/${fasta} -outDir sister -filter none
 

workflow.wdl

@@ -1,4 +1,5 @@
 import "./tasks/variantsNorm.wdl" as variantsNorm
+import "./tasks/sister.wdl" as sister
 import "./tasks/mendelian.wdl" as mendelian
 import "./tasks/zipIndex.wdl" as zipIndex
 import "./tasks/merge.wdl" as merge
@@ -48,14 +49,21 @@ workflow {{ project_name }} {
 			cluster_config=cluster_config,
 			disk_size=disk_size
 		}
-		call mendelian.mendelian as mendelian {
-			input: 
+		call sister.sister as sister{
+			input:
 			LCL5_vcf=LCL5variantsNorm.normed_vcf,
 			LCL6_vcf=LCL6variantsNorm.normed_vcf,
+			ref_dir=ref_dir,
+			fasta=fasta,
+			family_name=sample[8],
+			cluster_config=cluster_config,
+			disk_size=disk_size
+		}
+		call mendelian.mendelian as mendelian {
+			input: 
+			sister_vcf=sister.sister_consistent_vcf,
 			LCL7_vcf=LCL7variantsNorm.normed_vcf,
 			LCL8_vcf=LCL8variantsNorm.normed_vcf,
-			LCL5_name=sample[4],
-			LCL6_name=sample[5],
 			LCL7_name=sample[6],
 			LCL8_name=sample[7],
 			family_name=sample[8],