Merge pull request #43 from biowdl/somatic

fix somatic tasks and add supporting tasks

gatk.wdl

@@ -142,8 +142,8 @@ task CombineGVCFs {
              -V ~{sep=' -V ' gvcfFiles} \
              -L ~{sep=' -L ' intervals}
         else # TODO this should be handeled in wdl
-            ln -sf ~{select_first(gvcfFiles)} ~{outputPath}
-            ln -sf ~{select_first(gvcfFileIndexes)} ~{outputPath}.tbi
+            ln -sf ~{gvcfFiles[0]} ~{outputPath}
+            ln -sf ~{gvcfFileIndexes[0]} ~{outputPath}.tbi
         fi
     }
 
@@ -298,7 +298,7 @@ task MuTect2 {
         String? preCommand
 
         Array[File]+ inputBams
-        File inputBamIndex
+        Array[File]+ inputBamIndex
         File refFasta
         File refFastaIndex
         File refDict
@@ -331,6 +331,7 @@ task MuTect2 {
 
     output {
         File vcfFile = outputVcf
+        File vcfIndex = outputVcf + ".tbi"
     }
 
     runtime {

manta.wdl

+version 1.0
+
+task Somatic {
+    input {
+        File tumorBam
+        File tumorIndex
+        File? normalBam
+        File? normalIndex
+        File refFasta
+        File refFastaIndex
+        String runDir
+        File? callRegions
+        File? callRegionsIndex
+        Boolean exome = false
+        String? preCommand
+        String? installDir
+
+        Int cores = 1
+        Int memory = 4
+    }
+
+    String toolCommand = if defined(installDir)
+        then installDir + "bin/configMata.py"
+        else "configManta.py"
+
+    command {
+        set -e -o pipefail
+        ~{preCommand}
+        ~{toolCommand} \
+        ~{"--normalBam " + normalBam} \
+        ~{"--tumorBam " + tumorBam} \
+        --referenceFasta ~{refFasta} \
+        ~{"--callRegions " + callRegions} \
+        --runDir ~{runDir} \
+        ~{true="--exome" false="" exome}
+
+        ~{runDir}/runWorkflow.py \
+        -m local \
+        -j ~{cores} \
+        -g ~{memory}
+    }
+
+    output {
+        File condidateSmallIndels = runDir + "/results/variants/candidateSmallIndels.vcf.gz"
+        File condidateSmallIndelsIndex = runDir +
+            "/results/variants/candidateSmallIndels.vcf.gz.tbi"
+        File candidateSV = runDir + "/results/variants/candidateSV.vcf.gz"
+        File candidateSVindex = runDir + "/results/variants/candidateSV.vcf.gz.tbi"
+        File tumorSV = if defined(normalBam)
+            then runDir + "/results/variants/somaticSV.vcf.gz"
+            else runDir + "/results/variants/tumorSV.vcf.gz"
+        File tumorSVindex = if defined(normalBam)
+            then runDir + "/results/variants/somaticSV.vcf.gz.tbi"
+            else runDir + "/results/variants/tumorSV.vcf.gz.tbi"
+        File? diploidSV = "/results/variants/diploidSV.vcf.gz"
+        File? diploidSVindex = "/results/variants/diploidSV.vcf.gz.tbi"
+    }
+
+    runtime {
+        cpu: cores
+        memory: memory
+    }
+}
\ No newline at end of file

picard.wdl

@@ -378,3 +378,38 @@ task ScatterIntervalList {
         memory: ceil(memory * memoryMultiplier)
     }
 }
+
+task SortVcf {
+    input {
+        String? preCommand
+        String? picardJar
+
+        Array[File]+ vcfFiles
+        String outputVcf
+
+        Int memory = 4
+        Float memoryMultiplier = 3.0
+        }
+
+        String toolCommand = if defined(picardJar)
+            then "java -Xmx" + memory + "G -jar " + picardJar
+            else "picard -Xmx" + memory + "G"
+
+    command {
+        set -e -o pipefail
+        ~{preCommand}
+        ~{toolCommand} \
+        SortVcf \
+        I=~{sep=" I=" vcfFiles} \
+        O=outputVcf
+    }
+
+    output {
+        File vcfFile = outputVcf
+        File vcfIndex = outputVcf + ".tbi"
+    }
+
+    runtime {
+        memory: ceil(memory * memoryMultiplier)
+    }
+}
\ No newline at end of file

samtools.wdl

 version 1.0
 
+task BgzipAndIndex {
+    input {
+        File inputFile
+        String outputDir
+        String type = "vcf"
+    }
+
+    String outputGz = outputDir + "/" + basename(inputFile) + ".gz"
+
+    command {
+        bgzip -c ~{inputFile} > ~{outputGz}
+        tabix ~{outputGz} -p ~{type}
+    }
+
+    output {
+        File compressed = outputGz
+        File index = outputGz + ".tbi"
+    }
+}
+
 task Index {
     input {
         String? preCommand
@@ -131,6 +151,21 @@ task Fastq {
     }
 }
 
+task Tabix {
+    input {
+        String inputFile
+        String type = "vcf"
+    }
+
+    command {
+        tabix ~{inputFile} -p ~{type}
+    }
+
+    output {
+        File index = inputFile + ".tbi"
+    }
+}
+
 task View {
     input {
         String? preCommand
@@ -157,7 +192,7 @@ task View {
         ~{"-f " + includeFilter} \
         ~{"-F " + excludeFilter} \
         ~{"-G " + excludeSpecificFilter} \
-        ~{"--threads " + threads - 1} \
+        ~{"--threads " + (threads - 1)} \
         ~{inFile}
     }
 

strelka.wdl

 version 1.0
 
+task Germline {
+    input {
+        String? preCommand
+        String? installDir
+        String runDir
+        Array[File]+ bams
+        Array[File]+ indexes
+        File refFasta
+        File refFastaIndex
+        File? callRegions
+        File? callRegionsIndex
+        Boolean exome = false
+        Boolean rna = false
+
+        Int cores = 1
+        Int memory = 4
+    }
+
+    String toolCommand = if defined(installDir)
+        then installDir + "bin/configureStrelkaGermlineWorkflow.py"
+        else "configureStrelkaGermlineWorkflow.py"
+
+    command {
+        set -e -o pipefail
+        ~{preCommand}
+        ~{toolCommand} \
+        --bam ~{sep=" --bam " bams} \
+        --ref ~{refFasta} \
+        --runDir ~{runDir} \
+        ~{"--callRegions " + callRegions} \
+        ~{true="--exome" false="" exome} \
+        ~{true="--rna" false="" rna}
+
+        ~{runDir}/runWorkflow.py \
+        -m local \
+        -j ~{cores} \
+        -g ~{memory}
+    }
+
+    output {
+        File variants = runDir + "/results/variants/variants.vcf.gz"
+        File variantsIndex = runDir + "/results/variants/variants.vcf.gz.tbi"
+    }
+
+    runtime {
+        cpu: cores
+        memory: memory
+    }
+}
+
+
 task Somatic {
     input {
         String? preCommand
         String? installDir
         String runDir
         File normalBam
+        File normalIndex
         File tumorBam
+        File tumorIndex
         File refFasta
+        File refFastaIndex
+        File? callRegions
+        File? callRegionsIndex
+        File? indelCandidates
+        File? indelCandidatesIndex
+        Boolean exome = false
 
         Int cores = 1
         Int memory = 4
@@ -24,11 +83,14 @@ task Somatic {
         --normalBam ~{normalBam} \
         --tumorBam ~{tumorBam} \
         --ref ~{refFasta} \
-        --runDir ~{runDir}
+        --runDir ~{runDir} \
+        ~{"--callRegions " + callRegions} \
+        ~{"--indelCandidates " + indelCandidates} \
+        ~{true="--exome" false="" exome} \
 
         ~{runDir}/runWorkflow.py \
         -m local \
-        -J ~{cores} \
+        -j ~{cores} \
         -g ~{memory}
     }
 

vardict.wdl

@@ -3,13 +3,17 @@ version 1.0
 task VarDict {
     input {
         String? installDir
+        Boolean useJavaVersion = true
 
+        String tumorSampleName
         File tumorBam
-        File normalBam
+        File tumorIndex
+        String? normalSampleName
+        File? normalBam
+        File? normalIndex
         File refFasta
+        File refFastaIndex
         File bedFile
-        String tumorSampleName
-        String normalSampleName
         String outputVcf
 
         Int chromosomeColumn = 1
@@ -22,7 +26,9 @@ task VarDict {
 
     String toolCommand = if defined(installDir)
         then installDir + "/VarDict"
-        else "vardict"
+        else if useJavaVersion
+            then "vardict-java" #probably needs memory stuff
+            else "vardict"
 
     command {
         set -e -o pipefail
@@ -30,19 +36,21 @@ task VarDict {
         ~{toolCommand} \
         -G ~{refFasta} \
         -N ~{tumorSampleName} \
-        -b "~{tumorBam}|~{normalBam}" \
+        -b "~{tumorBam}~{"|" + normalBam}" \
         -c ~{chromosomeColumn} \
         -S ~{startColumn} \
         -E ~{endColumn} \
         -g ~{geneColumn} \
         ~{bedFile} | \
-        ~{installDir + "/"}testsomatic.R | \
-        ~{installDir + "/"}var2vcf_paired.pl \
-        -N "~{tumorSampleName}|~{normalSampleName}" \
-        > ~{outputVcf}
+        ~{installDir + "/"}~{true="testsomatic.R" false="teststrandbias.R" defined(normalBam)} | \
+        ~{installDir + "/"}~{true="var2vcf_paired.pl"
+            false="var2vcf_valid.pl" defined(normalBam)} \
+        -N "~{tumorSampleName}~{"|" + normalSampleName}" \
+        ~{true="" false="-E" defined(normalBam)} | \
+        bgzip -c > ~{outputVcf}
     }
 
     output {
         File vcfFile = outputVcf
     }
-}
\ No newline at end of file
+}