Merge branch 'master' of http://47.103.223.233/renluyao/quartet_dna_quality_control_big_pipeline

inputs

@@ -12,7 +12,7 @@
   "{{ project_name }}.FASTQCdocker": "{{ FASTQCdocker }}",
   "{{ project_name }}.MULTIQCdocker": "{{ MULTIQCdocker }}",
   "{{ project_name }}.fastq_2_M8": "{{ fastq_2_M8 }}",
-  "{{ project_name }}.project": "{{ project }}",
+  "{{ project_name }}.project": "{{ project_name }}",
   "{{ project_name }}.fastq_1_M8": "{{ fastq_1_M8 }}",
   "{{ project_name }}.SMALLcluster_config": "{{ SMALLcluster_config }}",
   "{{ project_name }}.screen_ref_dir": "{{ screen_ref_dir }}",

manifest.json

@@ -0,0 +1,21 @@
+{
+  "name": "WGS QC for Quartet",
+  "short_name": "quartet_dna_quality_control_wgs_big_pipeline",
+  "description": "Quality control of germline variants calling results using a Chinese Quartet family",
+  "home": "http://apps.clinico-omics.org/renluyao/quartet_dna_quality_control_wgs_big_pipeline",
+  "hidden": false,
+  "icons": [
+    {
+      "src": "",
+      "type": "image/png",
+      "sizes": "192x192"
+    },
+    {
+      "src": "",
+      "type": "image/png",
+      "sizes": "192x192"
+    }
+  ],
+  "category": "Pipeline",
+  "source": "PGx"
+}

schema.json

@@ -0,0 +1,307 @@
+{
+  "formMode": "batch",
+  "fields": [
+    {
+      "tmplType": "sample_id",
+      "type": "number",
+      "label": "Sample ID",
+      "name": "sample_id",
+      "question": "How many samples? For auto generating sample id.",
+      "placeholder": "How many samples? For auto generating sample id.",
+      "model": "sample_id",
+      "disabled": false,
+      "min": 1,
+      "max": 100,
+      "config": {
+        "initialValue": 1,
+        "rules": [
+          {
+            "type": "number",
+            "required": true,
+            "message": "Please input a number."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "radio",
+      "label": "FASTQ/VCF File?",
+      "model": "fastq_or_vcf",
+      "options": [
+        {
+          "label": "FASTQ",
+          "value": "fastq"
+        },
+        {
+          "label": "VCF",
+          "value": "vcf"
+        }
+      ],
+      "config": {
+        "rules": [
+          {
+            "type": "string",
+            "required": true,
+            "message": "Please select a file type."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D5 VCF Files",
+      "model": "vcf_D5",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "vcf",
+      "filterType": "vcf",
+      "placeholder": "Select files for D5 vcf.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D5 vcf."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D5 Read1 Files",
+      "model": "fastq_1_D5",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for D5 read1.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D5 read1."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D5 Read2 Files",
+      "model": "fastq_2_D5",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for D5 read2.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D5 read2."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D6 VCF Files",
+      "model": "vcf_D6",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "vcf",
+      "filterType": "vcf",
+      "placeholder": "Select files for D6 vcf.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D6 vcf."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D6 Read1 Files",
+      "model": "fastq_1_D6",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for D6 read1.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D6 read1."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "D6 Read2 Files",
+      "model": "fastq_2_D6",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for D6 read2.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for D6 read2."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "F7 VCF Files",
+      "model": "vcf_F7",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "vcf",
+      "filterType": "vcf",
+      "placeholder": "Select files for F7 vcf.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for F7 vcf."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "F7 Read1 Files",
+      "model": "fastq_1_F7",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for F7 read1.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for F7 read1."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "F7 Read2 Files",
+      "model": "fastq_2_F7",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for F7 read2.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for F7 read2."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "M8 VCF Files",
+      "model": "vcf_M8",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "vcf",
+      "filterType": "vcf",
+      "placeholder": "Select files for M8 vcf.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for M8 vcf."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "M8 Read1 Files",
+      "model": "fastq_1_M8",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for M8 read1.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for M8 read1."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "file",
+      "label": "M8 Read2 Files",
+      "model": "fastq_2_M8",
+      "readOnly": false,
+      "multiple": true,
+      "visibleVariable": "fastq_or_vcf",
+      "expectedValue": "fastq",
+      "filterType": "fastq|fastq.gz|fq|fq.gz",
+      "placeholder": "Select files for M8 read2.",
+      "config": {
+        "rules": [
+          {
+            "type": "array",
+            "required": true,
+            "message": "Please select files for M8 read2."
+          }
+        ]
+      }
+    },
+    {
+      "tmplType": "actions",
+      "buttons": [
+        {
+          "type": "cancel",
+          "buttonType": "default",
+          "buttonLabel": "Previous Step"
+        },
+        {
+          "type": "submit",
+          "buttonType": "primary",
+          "buttonLabel": "Next Step",
+          "validate": true
+        }
+      ]
+    }
+  ]
+}

tasks/benchmark.wdl

@@ -26,28 +26,28 @@ task benchmark {
 		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL8" > LCL8_name
 
 
-		if [[ ${sample} =~ "LCL5" ]];then
+		if [[ ${sample} =~ "D5" ]];then
 			/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
+		elif [[ ${sample} =~ "D6" ]]; then
 		    /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
+	    elif [[ ${sample} =~ "F7" ]]; then
 	        /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
+		elif [[ ${sample} =~ "M8" ]]; then
 			/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

tasks/fastqscreen.wdl

@@ -3,8 +3,6 @@ task fastq_screen {
 	File read2
 	File screen_ref_dir
 	File fastq_screen_conf
-	String read1name = basename(read1,".fastq.gz")
-	String read2name = basename(read2,".fastq.gz")
 	String docker
 	String cluster_config
 	String disk_size
@@ -15,22 +13,23 @@ task fastq_screen {
 		nt=$(nproc)
 		mkdir -p /cromwell_root/tmp
 		cp -r ${screen_ref_dir} /cromwell_root/tmp/
-		fastq_screen --aligner bowtie2 --conf ${fastq_screen_conf} --top 100000 --threads $nt ${read1}
-		fastq_screen --aligner bowtie2 --conf ${fastq_screen_conf} --top 100000 --threads $nt ${read2}
+		fastq_screen --aligner bowtie2 --conf ${fastq_screen_conf} --subset 1000000 --threads $nt ${read1}
+		fastq_screen --aligner bowtie2 --conf ${fastq_screen_conf} --subset 1000000 --threads $nt ${read2}
 	>>>
 
 	runtime {
 		docker:docker
-    	cluster: cluster_config
-    	systemDisk: "cloud_ssd 40"
-    	dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
 	}
+	
 	output {
-		File png1 = "${read1name}_screen.png"
-		File txt1 = "${read1name}_screen.txt"
-		File html1 = "${read1name}_screen.html"
-		File png2 = "${read2name}_screen.png"
-		File txt2 = "${read2name}_screen.txt"
-		File html2 = "${read2name}_screen.html"
+		File png1 = sub(basename(read1), "\\.(fastq|fq)\\.gz$", "_screen.png")
+		File txt1 = sub(basename(read1), "\\.(fastq|fq)\\.gz$", "_screen.txt")
+		File html1 = sub(basename(read1), "\\.(fastq|fq)\\.gz$", "_screen.html")
+		File png2 = sub(basename(read2), "\\.(fastq|fq)\\.gz$", "_screen.png")
+		File txt2 = sub(basename(read2), "\\.(fastq|fq)\\.gz$", "_screen.txt")
+		File html2 = sub(basename(read2), "\\.(fastq|fq)\\.gz$", "_screen.html")
 	}
 }