Convert R2 to Option[File]

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Star.scala

@@ -113,10 +113,10 @@ object Star {
    * @return Return Star
    *
    */
-  def apply(configurable: Configurable, R1: File, R2: File, outputDir: File, isIntermediate: Boolean = false, deps: List[File] = Nil): Star = {
+  def apply(configurable: Configurable, R1: File, R2: Option[File], outputDir: File, isIntermediate: Boolean = false, deps: List[File] = Nil): Star = {
     val star = new Star(configurable)
     star.R1 = R1
-    if (R2 != null) star.R2 = R2
+    R2.foreach(R2 => star.R2 = R2)
     star.outputDir = outputDir
     star.isIntermediate = isIntermediate
     star.deps = deps
@@ -136,11 +136,11 @@ object Star {
    */
   def _2pass(configurable: Configurable,
              R1: File,
-             R2: File,
+             R2: Option[File],
              outputDir: File,
              isIntermediate: Boolean = false,
              deps: List[File] = Nil): (File, List[Star]) = {
-    val starCommand_pass1 = Star(configurable, R1, if (R2 != null) R2 else null, new File(outputDir, "aln-pass1"))
+    val starCommand_pass1 = Star(configurable, R1, R2, new File(outputDir, "aln-pass1"))
     starCommand_pass1.isIntermediate = isIntermediate
     starCommand_pass1.deps = deps
     starCommand_pass1.beforeGraph
@@ -152,7 +152,7 @@ object Star {
     starCommand_reindex.isIntermediate = isIntermediate
     starCommand_reindex.beforeGraph
 
-    val starCommand_pass2 = Star(configurable, R1, if (R2 != null) R2 else null, new File(outputDir, "aln-pass2"))
+    val starCommand_pass2 = Star(configurable, R1, R2, new File(outputDir, "aln-pass2"))
     starCommand_pass2.genomeDir = starCommand_reindex.outputDir
     starCommand_pass2.isIntermediate = isIntermediate
     starCommand_pass2.deps = deps

public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala

@@ -95,7 +95,7 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
   def biopetScript() {
     runInitialJobs()
 
-    val out = if (paired) runTrimClip(outputFiles("fastq_input_R1"), outputFiles("fastq_input_R2"), outputDir)
+    val out = if (paired) runTrimClip(outputFiles("fastq_input_R1"), Some(outputFiles("fastq_input_R2")), outputDir)
     else runTrimClip(outputFiles("fastq_input_R1"), outputDir)
 
     val R1_files = for ((k, v) <- outputFiles if k.endsWith("output_R1")) yield v
@@ -130,8 +130,8 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
    * @param chunk
    * @return
    */
-  def runTrimClip(R1_in: File, outDir: File, chunk: String): (File, File, List[File]) =
-    runTrimClip(R1_in, new File(""), outDir, chunk)
+  def runTrimClip(R1_in: File, outDir: File, chunk: String): (File, Option[File], List[File]) =
+    runTrimClip(R1_in, None, outDir, chunk)
 
   /**
    * Adds all chunkable jobs of flexiprep
@@ -139,8 +139,8 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
    * @param outDir
    * @return
    */
-  def runTrimClip(R1_in: File, outDir: File): (File, File, List[File]) =
-    runTrimClip(R1_in, new File(""), outDir, "")
+  def runTrimClip(R1_in: File, outDir: File): (File, Option[File], List[File]) =
+    runTrimClip(R1_in, None, outDir, "")
 
   /**
    * Adds all chunkable jobs of flexiprep
@@ -149,7 +149,7 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
    * @param outDir
    * @return
    */
-  def runTrimClip(R1_in: File, R2_in: File, outDir: File): (File, File, List[File]) =
+  def runTrimClip(R1_in: File, R2_in: Option[File], outDir: File): (File, Option[File], List[File]) =
     runTrimClip(R1_in, R2_in, outDir, "")
 
   /**
@@ -160,14 +160,14 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
    * @param chunkarg
    * @return
    */
-  def runTrimClip(R1_in: File, R2_in: File, outDir: File, chunkarg: String): (File, File, List[File]) = {
+  def runTrimClip(R1_in: File, R2_in: Option[File], outDir: File, chunkarg: String): (File, Option[File], List[File]) = {
     val chunk = if (chunkarg.isEmpty || chunkarg.endsWith("_")) chunkarg else chunkarg + "_"
     var results: Map[String, File] = Map()
 
-    var R1: File = new File(R1_in)
-    var R2: File = if (paired) new File(R2_in) else null
-    var deps_R1: List[File] = R1 :: Nil
-    var deps_R2: List[File] = if (paired) R2 :: Nil else Nil
+    var R1 = R1_in
+    var R2 = R2_in
+    var deps_R1 = R1 :: Nil
+    var deps_R2 = if (paired) R2.get :: Nil else Nil
     def deps: List[File] = deps_R1 ::: deps_R2
 
     val seqtkSeq_R1 = SeqtkSeq(this, R1, swapExt(outDir, R1, R1_ext, ".sanger" + R1_ext), fastqc_R1)
@@ -178,12 +178,12 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
     deps_R1 ::= R1
 
     if (paired) {
-      val seqtkSeq_R2 = SeqtkSeq(this, R2, swapExt(outDir, R2, R2_ext, ".sanger" + R2_ext), fastqc_R2)
+      val seqtkSeq_R2 = SeqtkSeq(this, R2.get, swapExt(outDir, R2.get, R2_ext, ".sanger" + R2_ext), fastqc_R2)
       seqtkSeq_R2.isIntermediate = true
       add(seqtkSeq_R2)
       addSummarizable(seqtkSeq_R2, "seqtkSeq_R2")
-      R2 = seqtkSeq_R2.output
-      deps_R2 ::= R2
+      R2 = Some(seqtkSeq_R2.output)
+      deps_R2 ::= R2.get
     }
 
     val seqstat_R1 = Seqstat(this, R1, outDir)
@@ -193,7 +193,7 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
     addSummarizable(seqstat_R1, "seqstat_R1")
 
     if (paired) {
-      val seqstat_R2 = Seqstat(this, R2, outDir)
+      val seqstat_R2 = Seqstat(this, R2.get, outDir)
       seqstat_R2.isIntermediate = true
       seqstat_R2.deps = deps_R2
       add(seqstat_R2)
@@ -213,31 +213,31 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
       outputFiles += ("cutadapt_R1_stats" -> cutadapt_R1.stats_output)
 
       if (paired) {
-        val cutadapt_R2 = Cutadapt(this, R2, swapExt(outDir, R2, R2_ext, ".clip" + R2_ext))
+        val cutadapt_R2 = Cutadapt(this, R2.get, swapExt(outDir, R2.get, R2_ext, ".clip" + R2_ext))
         outputFiles += ("cutadapt_R2_stats" -> cutadapt_R2.stats_output)
         cutadapt_R2.fastqc = fastqc_R2
         cutadapt_R2.isIntermediate = true
         cutadapt_R2.deps = deps_R2
         add(cutadapt_R2)
         addSummarizable(cutadapt_R2, "clipping_R2")
-        R2 = cutadapt_R2.fastq_output
-        deps_R2 ::= R2
+        R2 = Some(cutadapt_R2.fastq_output)
+        deps_R2 ::= R2.get
 
         val fqSync = new FastqSync(this)
         fqSync.refFastq = cutadapt_R1.fastq_input
         fqSync.inputFastq1 = cutadapt_R1.fastq_output
         fqSync.inputFastq2 = cutadapt_R2.fastq_output
         fqSync.outputFastq1 = swapExt(outDir, R1, R1_ext, ".sync" + R1_ext)
-        fqSync.outputFastq2 = swapExt(outDir, R2, R2_ext, ".sync" + R2_ext)
+        fqSync.outputFastq2 = swapExt(outDir, R2.get, R2_ext, ".sync" + R2_ext)
         fqSync.outputStats = swapExt(outDir, R1, R1_ext, ".sync.stats")
         fqSync.deps :::= deps
         add(fqSync)
         addSummarizable(fqSync, "fastq_sync")
         outputFiles += ("syncStats" -> fqSync.outputStats)
         R1 = fqSync.outputFastq1
-        R2 = fqSync.outputFastq2
+        R2 = Some(fqSync.outputFastq2)
         deps_R1 ::= R1
-        deps_R2 ::= R2
+        deps_R2 ::= R2.get
       }
     }
 
@@ -246,9 +246,9 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
       sickle.input_R1 = R1
       sickle.output_R1 = swapExt(outDir, R1, R1_ext, ".trim" + R1_ext)
       if (paired) {
-        sickle.input_R2 = R2
-        sickle.output_R2 = swapExt(outDir, R2, R2_ext, ".trim" + R2_ext)
-        sickle.output_singles = swapExt(outDir, R2, R2_ext, ".trim.singles" + R1_ext)
+        sickle.input_R2 = R2.get
+        sickle.output_R2 = swapExt(outDir, R2.get, R2_ext, ".trim" + R2_ext)
+        sickle.output_singles = swapExt(outDir, R2.get, R2_ext, ".trim.singles" + R1_ext)
       }
       sickle.output_stats = swapExt(outDir, R1, R1_ext, ".trim.stats")
       sickle.deps = deps
@@ -256,7 +256,7 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
       add(sickle)
       addSummarizable(sickle, "trimming")
       R1 = sickle.output_R1
-      if (paired) R2 = sickle.output_R2
+      if (paired) R2 = Some(sickle.output_R2)
     }
 
     val seqstat_R1_after = Seqstat(this, R1, outDir)
@@ -265,14 +265,14 @@ class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with
     addSummarizable(seqstat_R1_after, "seqstat_R1_after")
 
     if (paired) {
-      val seqstat_R2_after = Seqstat(this, R2, outDir)
+      val seqstat_R2_after = Seqstat(this, R2.get, outDir)
       seqstat_R2_after.deps = deps_R2
       add(seqstat_R2_after)
       addSummarizable(seqstat_R2_after, "seqstat_R2_after")
     }
 
     outputFiles += (chunk + "output_R1" -> R1)
-    if (paired) outputFiles += (chunk + "output_R2" -> R2)
+    if (paired) outputFiles += (chunk + "output_R2" -> R2.get)
     return (R1, R2, deps)
   }
 

public/mapping/src/main/scala/nl/lumc/sasc/biopet/pipelines/mapping/Mapping.scala

@@ -157,26 +157,27 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
     var fastq_R1_output: List[File] = Nil
     var fastq_R2_output: List[File] = Nil
 
-    def removeGz(file: String): String = {
-      if (file.endsWith(".gz")) file.substring(0, file.lastIndexOf(".gz"))
-      else if (file.endsWith(".gzip")) file.substring(0, file.lastIndexOf(".gzip"))
+    def removeGz(file: File): File = {
+      val absPath = file.getAbsolutePath
+      if (absPath.endsWith(".gz")) new File(absPath.substring(0, absPath.lastIndexOf(".gz")))
+      else if (absPath.endsWith(".gzip")) new File(absPath.substring(0, absPath.lastIndexOf(".gzip")))
       else file
     }
-    var chunks: Map[File, (String, String)] = Map()
-    if (chunking) for (t <- 1 to numberChunks.getOrElse(1)) {
-      val chunkDir = new File(outputDir, "chunks" + File.separator + t)
-      chunks += (chunkDir -> (removeGz(chunkDir + input_R1.getName),
-        if (paired) removeGz(chunkDir + input_R2.get.getName) else ""))
+
+    val chunks: Map[File, (File, Option[File])] = {
+      if (chunking) {
+        (for (t <- 1 to numberChunks.getOrElse(1)) yield {
+          val chunkDir = new File(outputDir, "chunks" + File.separator + t)
+          (chunkDir -> (removeGz(new File(chunkDir, input_R1.getName)),
+            if (paired) Some(removeGz(new File(chunkDir, input_R2.get.getName))) else None))
+        }).toMap
+      } else if (skipFlexiprep) {
+        Map(outputDir -> (
+          extractIfNeeded(input_R1, flexiprep.outputDir),
+          if (paired) Some(extractIfNeeded(input_R2.get, outputDir)) else None)
+        )
+      } else Map(outputDir -> (input_R1, input_R2))
     }
-    else if (skipFlexiprep) {
-      chunks += (outputDir -> (
-        extractIfNeeded(input_R1, flexiprep.outputDir),
-        if (paired) extractIfNeeded(input_R2.get, outputDir) else "")
-      )
-    } else chunks += (outputDir -> (
-      flexiprep.outputFiles("fastq_input_R1"),
-      if (paired) flexiprep.outputFiles("fastq_input_R2") else "")
-    )
 
     if (chunking) {
       val fastSplitter_R1 = new FastqSplitter(this)
@@ -188,7 +189,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
       if (paired) {
         val fastSplitter_R2 = new FastqSplitter(this)
         fastSplitter_R2.input = input_R2.get
-        for ((chunkDir, fastqfile) <- chunks) fastSplitter_R2.output :+= fastqfile._2
+        for ((chunkDir, fastqfile) <- chunks) fastSplitter_R2.output :+= fastqfile._2.get
         fastSplitter_R2.isIntermediate = true
         add(fastSplitter_R2)
       }
@@ -205,7 +206,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
         if (paired) R2 = flexiout._2
         deps = flexiout._3
         fastq_R1_output :+= R1
-        fastq_R2_output :+= R2
+        R2.foreach(R2 => fastq_R2_output :+= R2)
       }
 
       val outputBam = new File(chunkDir, outputName + ".bam")
@@ -267,7 +268,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addBwaAln(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addBwaAln(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
     val bwaAlnR1 = new BwaAln(this)
     bwaAlnR1.fastq = R1
     bwaAlnR1.deps = deps
@@ -277,7 +278,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
 
     val samFile: File = if (paired) {
       val bwaAlnR2 = new BwaAln(this)
-      bwaAlnR2.fastq = R2
+      bwaAlnR2.fastq = R2.get
       bwaAlnR2.deps = deps
       bwaAlnR2.output = swapExt(output.getParent, output, ".bam", ".R2.sai")
       bwaAlnR2.isIntermediate = true
@@ -285,7 +286,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
 
       val bwaSampe = new BwaSampe(this)
       bwaSampe.fastqR1 = R1
-      bwaSampe.fastqR2 = R2
+      bwaSampe.fastqR2 = R2.get
       bwaSampe.saiR1 = bwaAlnR1.output
       bwaSampe.saiR2 = bwaAlnR2.output
       bwaSampe.r = getReadGroupBwa
@@ -320,10 +321,10 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addBwaMem(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addBwaMem(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
     val bwaCommand = new BwaMem(this)
     bwaCommand.R1 = R1
-    if (paired) bwaCommand.R2 = R2
+    if (paired) bwaCommand.R2 = R2.get
     bwaCommand.deps = deps
     bwaCommand.R = Some(getReadGroupBwa)
     bwaCommand.output = swapExt(output.getParent, output, ".bam", ".sam")
@@ -335,10 +336,9 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
     sortSam.output
   }
 
-  def addGsnap(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addGsnap(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
     val gsnapCommand = new Gsnap(this)
-    // FIXME: ideally we should only check for null ~ but apparently it's possible to get a File object with "" as the path
-    gsnapCommand.input = List(R1, R2).filterNot(x => x == null || x.getPath == "")
+    gsnapCommand.input = if (paired) List(R1, R2.get) else List(R1)
     gsnapCommand.deps = deps
     gsnapCommand.output = swapExt(output.getParent, output, ".bam", ".sam")
     gsnapCommand.isIntermediate = true
@@ -360,12 +360,11 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
     addAddOrReplaceReadGroups(reorderSam.output, output)
   }
 
-  def addTophat(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addTophat(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
     // TODO: merge mapped and unmapped BAM ~ also dealing with validation errors in the unmapped BAM
     val tophat = new Tophat(this)
-    // FIXME: ideally we should only check for null ~ but apparently it's possible to get a File object with "" as the path
     tophat.R1 = tophat.R1 :+ R1
-    if (R2 != null && R2.getPath != "") tophat.R2 = tophat.R2 :+ R2
+    if (paired) tophat.R2 = tophat.R2 :+ R2.get
     tophat.output_dir = new File(outputDir, "tophat_out")
     tophat.deps = deps
     // always output BAM
@@ -376,7 +375,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
 
     val sortSam = new SortSam(this)
     sortSam.input = tophat.outputAcceptedHits
-    sortSam.output = swapExt(output.getParent, output, ".bam", ".sorted.bam")
+    sortSam.output = swapExt(output.getParentFile, output, ".bam", ".sorted.bam")
     sortSam.sortOrder = "coordinate"
     sortSam.isIntermediate = chunking || !skipMarkduplicates
     add(sortSam)
@@ -391,7 +390,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addStampy(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addStampy(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
 
     var RG: String = "ID:" + readgroupId + ","
     RG += "SM:" + sampleId.get + ","
@@ -405,7 +404,7 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
 
     val stampyCmd = new Stampy(this)
     stampyCmd.R1 = R1
-    if (paired) stampyCmd.R2 = R2
+    if (paired) stampyCmd.R2 = R2.get
     stampyCmd.deps = deps
     stampyCmd.readgroup = RG
     stampyCmd.sanger = true
@@ -426,10 +425,10 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addBowtie(R1: File, R2: File, output: File, deps: List[File]): File = {
+  def addBowtie(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
     val bowtie = new Bowtie(this)
     bowtie.R1 = R1
-    if (paired) bowtie.R2 = Some(R2)
+    if (paired) bowtie.R2 = R2
     bowtie.deps = deps
     bowtie.output = this.swapExt(output.getParent, output, ".bam", ".sam")
     bowtie.isIntermediate = true
@@ -445,8 +444,8 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addStar(R1: File, R2: File, output: File, deps: List[File]): File = {
-    val starCommand = Star(this, R1, if (paired) R2 else null, outputDir, isIntermediate = true, deps = deps)
+  def addStar(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
+    val starCommand = Star(this, R1, R2, outputDir, isIntermediate = true, deps = deps)
     add(starCommand)
     addAddOrReplaceReadGroups(starCommand.outputSam, output)
   }
@@ -459,8 +458,8 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
    * @param deps
    * @return
    */
-  def addStar2pass(R1: File, R2: File, output: File, deps: List[File]): File = {
-    val starCommand = Star._2pass(this, R1, if (paired) R2 else null, outputDir, isIntermediate = true, deps = deps)
+  def addStar2pass(R1: File, R2: Option[File], output: File, deps: List[File]): File = {
+    val starCommand = Star._2pass(this, R1, R2, outputDir, isIntermediate = true, deps = deps)
     addAll(starCommand._2)
     addAddOrReplaceReadGroups(starCommand._1, output)
   }