Merge branch 'develop' into BIOWDL-85

biopet.wdl

@@ -409,6 +409,8 @@ task ValidateFastq {
 
     output {
         File stderr = stderr()
+        File validatedFastq1 = fastq1
+        File? validatedFastq2 = fastq2
     }
 
     runtime {

biopet/seqstat.wdl

+version 1.0
+
+# Copyright Sequencing Analysis Support Core - Leiden University Medical Center 2018
+
+task Generate {
+    input {
+        String? preCommand
+        File? toolJar
+        File fastqR1
+        File? fastqR2
+        String outputFile
+        String sample
+        String library
+        String readgroup
+
+        Int memory = 4
+        Float memoryMultiplier = 2.0
+    }
+
+    String toolCommand = if defined(toolJar)
+        then "java -Xmx" + memory + "G -jar " + toolJar
+        else "biopet-seqstat -Xmx" + memory + "G"
+
+    command {
+        set -e -o pipefail
+        ~{preCommand}
+        mkdir -p $(dirname ~{outputFile})
+        ~{toolCommand} Generate \
+        --fastqR1 ~{fastqR1} \
+        ~{"--fastqR2 " + fastqR2} \
+        --output ~{outputFile} \
+        ~{"--sample " + sample} \
+        ~{"--library " + library } \
+        ~{"--readgroup " + readgroup }
+    }
+
+    output {
+        File json = outputFile
+    }
+
+    runtime {
+        memory: ceil(memory * memoryMultiplier)
+    }
+}
\ No newline at end of file

bwa.wdl

 version 1.0
 
+import "common.wdl" as common
+
 task Mem {
     input {
         String? preCommand
@@ -9,29 +11,59 @@ task Mem {
         String outputPath
         String? readgroup
 
+        String? picardJar
+
         Int threads = 1
         Int memory = 8
+        Int picardMemory = 4
     }
 
+    String picardPrefix = if defined(picardJar)
+        then "java -Xmx" + picardMemory + "G -jar " + picardJar
+        else "picard -Xmx" + picardMemory + "G"
+
+    # Post alt script from bwa
+    String altCommand = if (defined(bwaIndex.altIndex)) then "| bwa-postalt " + bwaIndex.altIndex else ""
+
+    # setNmMdAndUqTags is only required if alt sequences are added
+    String setNmMdAndUqTagsCommand = picardPrefix + " SetNmMdAndUqTags " +
+                                             " INPUT=/dev/stdin OUTPUT=" + outputPath +
+                                             " CREATE_INDEX=true" +
+                                             " R=" + bwaIndex.fastaFile
+
+    String sortSamCommand = picardPrefix + " SortSam " +
+                 " INPUT=/dev/stdin SORT_ORDER=coordinate " +
+                 if(defined(bwaIndex.altIndex)) then " OUTPUT=/dev/stdout "
+                 else " OUTPUT=" + outputPath + " CREATE_INDEX=true "
+
+    String picardCommand = if (defined(bwaIndex.altIndex)) then sortSamCommand + " | " + setNmMdAndUqTagsCommand
+    else sortSamCommand
+
+    String readgroupArg = if (defined(readgroup)) then "-R '" + readgroup + "'" else ""
+
     command {
         set -e -o pipefail
         mkdir -p $(dirname ~{outputPath})
         ~{preCommand}
         bwa mem ~{"-t " + threads} \
-        ~{"-R '" + readgroup + "'"} \
+        ~{readgroupArg} \
         ~{bwaIndex.fastaFile} \
         ~{inputR1} \
         ~{inputR2} \
-        | samtools sort --output-fmt BAM - > ~{outputPath}
+        ~{altCommand} \
+        | ~{picardCommand}
     }
 
     output {
-        File bamFile = outputPath
+        IndexedBamFile bamFile = object {
+          file: outputPath,
+          index: sub(outputPath, ".bam$", ".bai")
+        }
     }
 
     runtime{
         cpu: threads
-        memory: memory
+        memory: memory + picardMemory + picardMemory
     }
 }
 
@@ -62,8 +94,10 @@ task Index {
     }
 
     output {
-        File indexedFasta = outputFile
-        Array[File] indexFiles = [outputFile + ".bwt",outputFile + ".pac",outputFile + ".sa",outputFile + ".amb",outputFile + ".ann"]
+        BwaIndex outputIndex = object {
+            fastaFile: outputFile,
+            indexFiles: [outputFile + ".bwt",outputFile + ".pac",outputFile + ".sa",outputFile + ".amb",outputFile + ".ann"]
+        }
     }
 
     parameter_meta {
@@ -77,4 +111,5 @@ task Index {
 struct BwaIndex {
     File fastaFile
     Array[File] indexFiles
+    File? altIndex
 }

centrifuge.wdl

@@ -101,7 +101,7 @@ task Classify {
         ~{true="-k" false="" defined(assignments)} ~{assignments} \
         ~{true="-1" false="-U" defined(read2)} ~{sep=',' read1} \
         ~{true="-2" false="" defined(read2)} ~{sep=',' read2} \
-        ~{true="-U" false="" defined(unpairedReads)} ~{sep=',' unpairedReads} \
+        ~{true="-U" false="" length(select_first([unpairedReads])) > 0} ~{sep=',' unpairedReads} \
         ~{"--report-file " + reportFilePath} \
         ~{"--min-hitlen " + minHitLen} \
         ~{"--min-totallen " + minTotalLen} \
@@ -197,7 +197,7 @@ task Kreport {
         String suffix = "kreport"
         String prefix = "centrifuge"
         String indexPrefix
-        Boolean? onlyUnique
+        Boolean? onlyUnique ## removed in 1.0.4
         Boolean? showZeros
         Boolean? isCountTable
         Int? minScore

common.wdl

@@ -42,11 +42,14 @@ task ConcatenateTextFiles {
         Boolean zip = false
     }
 
+    # When input and output is both compressed decompression is not needed
+    String cmdPrefix = if (unzip && !zip) then "zcat " else "cat "
+    String cmdSuffix = if (!unzip && zip) then " | gzip -c " else ""
+
     command {
         set -e -o pipefail
         ~{"mkdir -p $(dirname " + combinedFilePath + ")"}
-        ~{true='zcat' false= 'cat' unzip} ~{sep=' ' fileList} \
-        ~{true="| gzip -c" false="" zip} > ~{combinedFilePath}
+        ~{cmdPrefix} ~{sep=' ' fileList} ~{cmdSuffix} > ~{combinedFilePath}
     }
 
     output {
@@ -150,3 +153,10 @@ struct IndexedBamFile {
     File file
     File index
 }
+
+struct FastqPair {
+    File R1
+    String? R1_md5
+    File? R2
+    String? R2_md5
+}

gatk.wdl

@@ -12,7 +12,6 @@ task ApplyBQSR {
         File recalibrationReport
         Array[File]+ sequenceGroupInterval
         Reference reference
-        Int? compressionLevel
 
         Int memory = 4
         Float memoryMultiplier = 3.0
@@ -112,7 +111,6 @@ task CombineGVCFs {
 
         Reference reference
 
-        Int? compressionLevel #TODO This isn't being used?
         Int memory = 4
         Float memoryMultiplier = 3.0
     }
@@ -198,7 +196,6 @@ task GenotypeGVCFs {
 
         IndexedVcfFile? dbsnpVCF
 
-        Int? compressionLevel
         Int memory = 4
         Float memoryMultiplier =3.0
     }
@@ -243,7 +240,6 @@ task HaplotypeCallerGvcf {
         String gvcfPath
         Reference reference
         Float contamination = 0.0
-        Int? compressionLevel
         String? gatkJar
 
         IndexedVcfFile? dbsnpVCF

picard.wdl

 version 1.0
 
+task BedToIntervalList {
+    input {
+        String? preCommand
+        File? picardJar
+
+        File bedFile
+        File dict
+        String outputPath
+
+        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
+        mkdir -p $(dirname "~{outputPath}")
+        ~{preCommand}
+        ~{toolCommand} \
+        BedToIntervalList \
+        I=~{bedFile} \
+        O=~{outputPath} \
+        SD=~{dict}
+    }
+
+    output {
+        File intervalList = outputPath
+    }
+
+    runtime {
+        memory: ceil(memory * memoryMultiplier)
+    }
+}
+
 task CollectMultipleMetrics {
     input {
         String? preCommand

seqtk.wdl

@@ -6,9 +6,8 @@ task Sample {
         String outFilePath = "subsampledReads.fq.gz"
         String? preCommand
         Int? seed
-        Boolean twoPassMode
-        Float? fraction
-        Int? number
+        Boolean twoPassMode = false
+        Float fractionOrNumber # when above 1.0 is the number of reads, otherwise it's a fraction
         Boolean zip = true
     }
 
@@ -20,8 +19,7 @@ task Sample {
         ~{"-s " + seed} \
         ~{true="-2 " false="" twoPassMode} \
         ~{sequenceFile} \
-        ~{number} \
-        ~{fraction} \
+        ~{fractionOrNumber} \
         ~{true="| gzip" false="" zip} \
         >  ~{outFilePath}
     }

spades.wdl

@@ -32,7 +32,8 @@ task Spades {
         Int? phredOffset
     }
 
-    Int clusterMemory = ceil(memoryGb / threads)
+    Int clusterMemory = ceil(memoryGb / threads * 1.2)
+    Int memoryArg = ceil(memoryGb)
 
     command {
         set -e -o pipefail
@@ -60,7 +61,7 @@ task Spades {
         ~{true="--disable-rr" false="" disableRepeatResolution} \
         ~{"--dataset " + dataset} \
         ~{"--threads " + threads} \
-        ~{"--memory " + memoryGb} \
+        ~{"--memory " + memoryArg} \
         ~{"-k " + k} \
         ~{"--cov-cutoff " + covCutoff} \
         ~{"--phred-offset " + phredOffset}

star.wdl

@@ -20,7 +20,7 @@ task Star {
         Int memory = 10
     }
 
-    #TODO needs to be extended for all possible output extensions
+    # Needs to be extended for all possible output extensions
     Map[String, String] samOutputNames = {"BAM SortedByCoordinate": "sortedByCoord.out.bam"}
 
     command {