first commit coverageProfiler wdl

CoverageProfiler/CoverageProfiler.inputs.json

@@ -0,0 +1,12 @@
+{
+  "coverageProfile.referenceFasta": "File",
+  "coverageProfile.DepthOfCoverage.gatk_docker": "String (optional, default = \"broadinstitute/gatk:4.5.0.0\")",
+  "coverageProfile.intervals": "File",
+  "coverageProfile.alignedBam": "File",
+  "coverageProfile.DepthOfCoverage.command_mem_mb": "Int? (optional)",
+  "coverageProfile.DepthOfCoverage.cpu": "Int? (optional)",
+  "coverageProfile.DepthOfCoverage.machine_mem_mb": "Int? (optional)",
+  "coverageProfile.DepthOfCoverage.mem_gb": "Int? (optional)",
+  "coverageProfile.sampleName": "String"
+}
+

CoverageProfiler/CoverageProfiler.wdl

@@ -0,0 +1,60 @@
+version 1.0
+
+workflow coverageProfile {
+    input {
+        String sampleName
+        File alignedBam
+        File referenceFasta
+        File intervals
+    }
+    call DepthOfCoverage {
+        input:
+            sampleName = sampleName,
+            alignedBam = alignedBam,
+            referenceFasta = referenceFasta,
+            intervals = intervals
+    }
+    output {
+        File coveragebyInterval = DepthOfCoverage.sample_interval_summary
+        Float meanCoverage = DepthOfCoverage.mean_coverage
+    }
+    meta {
+        author: "Yueyao Gao"
+        email: "[email protected]"
+        description: "Calculates the depth of coverage of an input sample using GATK's DepthOfCoverage tool."
+
+    }
+}
+    task DepthOfCoverage {
+    input {
+        String sampleName
+        File alignedBam
+        File referenceFasta
+        File intervals
+        Int? mem_gb
+        Int? cpu
+        Int machine_mem_mb = select_first([mem_gb, 7]) * 1000
+        Int command_mem_mb = machine_mem_mb - 1000
+        String gatk_docker = "broadinstitute/gatk:4.5.0.0"
+    }
+    command <<<
+        mkdir output
+        gatk --java-options "-Xmx~{command_mem_mb}m" DepthOfCoverage \
+            -L ~{intervals} \
+            --input ~{alignedBam} \
+            --reference ~{referenceFasta} \
+            --output output/~{sampleName}
+
+        cat output/~{sampleName}.sample_interval_summary | awk 'BEGIN {FS = ","}{print $3}' | tail -n 1 > output/mean_coverage.txt
+    >>>
+    output {
+        File sample_interval_summary = "output/${sampleName}.sample_interval_summary"
+        Float mean_coverage = read_float("output/mean_coverage.txt")
+    }
+    runtime {
+        memory: machine_mem_mb + " MB"
+        cpu: select_first([cpu, 1])
+        docker: gatk_docker
+        disks: "local-disk 500 HDD"
+    }
+}