first commit

README.md

@@ -0,0 +1,43 @@
+#突变的数目统计
+
+> Author： Run Luyao
+> 
+> E-mail：18110700050@fudan.edu.cn
+> 
+> Git: http://choppy.3steps.cn/renluyao/VCFstat.git
+> 
+> Last Updates: 2020/03/02
+
+## 安装指南
+```
+# 激活choppy环境
+source activate choppy
+# 安装app
+choppy install renluyao/VCFstat
+```
+
+## App概述
+
+用于统计VCF的突变数目等基本情况。用的软件是RTGtools vcfstats
+
+
+## 流程与参数
+
+
+## App输入变量与输入文件
+
+
+
+## App输出文件
+
+
+
+
+## 结果展示与解读
+
+
+
+## CHANGELOG
+
+
+## FAQ

inputs

@@ -0,0 +1,9 @@
+{
+  "{{ project_name }}.vcfstat.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-hap:latest",
+  "{{ project_name }}.vcfstat.disk_size": "100",
+  "{{ project_name }}.inputSamplesFile": "{{ inputSamplesFile }}",
+  "{{ project_name }}.mergeNum.docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/gatk:v2019.01",
+  "{{ project_name }}.mergeNum.disk_size": "100",
+  "{{ project_name }}.mergeNum.cluster_config": "OnDemand bcs.a2.large img-ubuntu-vpc",
+  "{{ project_name }}.vcfstat.cluster_config": "OnDemand bcs.a2.large img-ubuntu-vpc"
+}

tasks/mergeNum.wdl

@@ -0,0 +1,26 @@
+task mergeNum {
+	Array[File] vcfnumber
+	String docker
+	String cluster_config
+	String disk_size
+
+	command <<<
+		set -o pipefail
+		set -e
+		for i in ${sep=" " vcfnumber}
+		do
+		  cat $i | cut -d':' -f2 | tr '\n' '\t' | sed s'/\t$/\n/g' >> vcfstats
+		done
+		sed '1i\File\tFailed Filters\tPassed Filters\tSNPs\tMNPs\tInsertions\tDeletions\tIndels\tSame as reference\tSNP Transitions/Transversions\tTotal Het/Hom ratio\tSNP Het/Hom ratio\tMNP Het/Hom ratio\tInsertion Het/Hom ratio\tDeletion Het/Hom ratio\tIndel Het/Hom ratio\tInsertion/Deletion ratio\tIndel/SNP+MNP ratio' vcfstats > vcfstats.txt
+	>>>
+
+	runtime {
+		docker:docker
+    	cluster:cluster_config
+    	systemDisk:"cloud_ssd 40"
+    	dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File vcfstat="vcfstats.txt"
+	}
+}

tasks/vcfstat.wdl

@@ -0,0 +1,25 @@
+task vcfstat {
+	File vcf
+	String sampleName
+	String docker
+	String cluster_config
+	String disk_size
+
+	command <<<
+		set -o pipefail
+		set -e
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${vcf} -c > ${sampleName}.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sampleName}.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfstats ${sampleName}.vcf.gz > ${sampleName}_onestats.txt
+	>>>
+
+	runtime {
+		docker:docker
+    	cluster:cluster_config
+    	systemDisk:"cloud_ssd 40"
+    	dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+	output {
+		File vcfnumber="${sampleName}_onestats.txt"
+	}
+}

workflow.wdl

@@ -0,0 +1,25 @@
+import "./tasks/vcfstat.wdl" as vcfstat
+import "./tasks/mergeNum.wdl" as mergeNum
+
+workflow {{ project_name }} {
+
+	File inputSamplesFile
+	Array[Array[File]] inputSamples = read_tsv(inputSamplesFile)
+	
+
+	scatter (sample in inputSamples) {
+
+		call vcfstat.vcfstat as vcfstat {
+			input:
+			vcf=sample[1],
+			sampleName=sample[0]
+		}	
+
+	}
+
+	call mergeNum.mergeNum as mergeNum {
+		input:
+		vcfnumber=vcfstat.vcfnumber
+	}
+
+}