Bugfix in yamsvp (concerns multiple library merging), Adding Sambamba tool wrappers

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/sambamba/Sambamba.scala

+package nl.lumc.sasc.biopet.extensions.sambamba
+
+import nl.lumc.sasc.biopet.core.BiopetCommandLineFunction
+
+abstract class Sambamba extends BiopetCommandLineFunction {
+  override val defaultVmem = "4G"
+  override val defaultThreads = 2
+
+  executable = config("exe", default = "sambamba", submodule = "sambamba", freeVar = false)
+  override def versionCommand = executable
+  override val versionRegex = """sambamba v(.*)""".r
+  override val versionExitcode = List(0, 1)
+}
\ No newline at end of file

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/sambamba/SambambaFlagstat.scala

+package nl.lumc.sasc.biopet.extensions.sambamba
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
+import java.io.File
+
+class SambambaFlagstat(val root: Configurable) extends Sambamba {
+  override val defaultThreads = 2
+
+  @Input(doc = "Bam File")
+  var input: File = _
+
+  @Output(doc = "output File")
+  var output: File = _
+
+  def cmdLine = required(executable) + 
+  required("flagstat") + 
+  optional("-t", nCoresRequest) +
+  required(input) + 
+  " > " + 
+  required(output)
+}
+
+object SambambaFlagstat {
+  def apply(root: Configurable, input: File, output: File): SambambaFlagstat = {
+    val flagstat = new SambambaFlagstat(root)
+    flagstat.input = input
+    flagstat.output = output
+    return flagstat
+  }
+
+  def apply(root: Configurable, input: File, outputDir: String): SambambaFlagstat = {
+    val dir = if (outputDir.endsWith("/")) outputDir else outputDir + "/"
+    val outputFile = new File(dir + swapExtension(input.getName))
+    return apply(root, input, outputFile)
+  }
+
+  def apply(root: Configurable, input: File): SambambaFlagstat = {
+    return apply(root, input, new File(swapExtension(input.getAbsolutePath)))
+  }
+
+  private def swapExtension(inputFile: String) = inputFile.stripSuffix(".bam") + ".flagstat"
+}
\ No newline at end of file

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/sambamba/SambambaIndex.scala

+package nl.lumc.sasc.biopet.extensions.sambamba
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
+import java.io.File
+
+class SambambaIndex(val root: Configurable) extends Sambamba {
+  override val defaultThreads = 2
+
+  @Input(doc = "Bam File")
+  var input: File = _
+
+  @Output(doc = "Output .bai file to")
+  var output: File = _
+
+  def cmdLine = required(executable) + 
+  required("index") + 
+  optional("-t", nCoresRequest) +
+  required(input) + 
+  required(output)
+}
+
+object SambambaIndex {
+  def apply(root: Configurable, input: File, output: File): SambambaIndex = {
+    val flagstat = new SambambaIndex(root)
+    flagstat.input = input
+    flagstat.output = output
+    return flagstat
+  }
+
+  def apply(root: Configurable, input: File, outputDir: String): SambambaIndex = {
+    val dir = if (outputDir.endsWith("/")) outputDir else outputDir + "/"
+    val outputFile = new File(dir + swapExtension(input.getName))
+    return apply(root, input, outputFile)
+  }
+
+  def apply(root: Configurable, input: File): SambambaIndex = {
+    return apply(root, input, new File(swapExtension(input.getAbsolutePath)))
+  }
+
+  private def swapExtension(inputFile: String) = inputFile.stripSuffix(".bam") + ".bam.bai"
+}
\ No newline at end of file

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/sambamba/SambambaMerge.scala

+package nl.lumc.sasc.biopet.extensions.sambamba
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
+import java.io.File
+
+class SambambaMerge(val root: Configurable) extends Sambamba {
+  override val defaultThreads = 4
+
+  @Input(doc = "Bam File[s]")
+  var input: List[File] = Nil
+
+  @Output(doc = "Output merged bam PATH")
+  var output: File = _
+
+  // @doc: compression_level 6 is average, 0 = no compression, 9 = best
+  val compression_level: Option[Int] = config("compression_level", default=6) 
+  
+  def cmdLine = required(executable) + 
+  required("merge") + 
+  optional("-t", nCoresRequest) +
+  optional("-l", compression_level) +
+  required(output) + 
+  repeat("", input)
+}
+
+object SambambaMerge {
+  def apply(root: Configurable, input: List[File], output: File): SambambaMerge = {
+    val flagstat = new SambambaMerge(root)
+    flagstat.input = input
+    flagstat.output = output
+    return flagstat
+  }
+
+  def apply(root: Configurable, input: List[File], outputDir: String): SambambaMerge = {
+    val dir = if (outputDir.endsWith("/")) outputDir else outputDir + "/"
+    val outputFile = new File(dir + swapExtension(input.head.getName))
+    return apply(root, input, outputFile)
+  }
+
+  def apply(root: Configurable, input: List[File]): SambambaMerge = {
+    return apply(root, input, new File(swapExtension(input.head.getAbsolutePath)))
+  }
+
+  private def swapExtension(inputFile: String) = inputFile.stripSuffix(".bam") + ".merge.bam"
+}
+
+//
+//object MergeSamFiles {
+//  def apply(root: Configurable, input: List[File], outputDir: String, sortOrder: String = null): MergeSamFiles = {
+//    val mergeSamFiles = new MergeSamFiles(root)
+//    mergeSamFiles.input = input
+//    mergeSamFiles.output = new File(outputDir, input.head.getName.stripSuffix(".bam").stripSuffix(".sam") + ".merge.bam")
+//    if (sortOrder == null) mergeSamFiles.sortOrder = "coordinate"
+//    else mergeSamFiles.sortOrder = sortOrder
+//    return mergeSamFiles
+//  }
+//}
\ No newline at end of file

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/pipelines/yamsvp/Yamsvp.scala

@@ -9,8 +9,10 @@ import nl.lumc.sasc.biopet.core.MultiSampleQScript
 import nl.lumc.sasc.biopet.core.PipelineCommand
 
 import nl.lumc.sasc.biopet.extensions.picard.MergeSamFiles
-import nl.lumc.sasc.biopet.extensions.svcallers.Breakdancer
-import nl.lumc.sasc.biopet.extensions.svcallers.Clever
+import nl.lumc.sasc.biopet.extensions.sambamba.{ SambambaIndex, SambambaMerge }
+import nl.lumc.sasc.biopet.extensions.svcallers.pindel.Pindel
+import nl.lumc.sasc.biopet.extensions.svcallers.{ Breakdancer, Clever }
+
 
 import nl.lumc.sasc.biopet.pipelines.mapping.Mapping
 
@@ -54,7 +56,7 @@ class Yamsvp(val root: Configurable) extends QScript with MultiSampleQScript {
   }
 
   override def onExecutionDone(jobs: Map[QFunction, JobRunInfo], success: Boolean) {
-    logger.info("YAM SV Pipeline has run ...............................................................")
+    logger.info("YAM SV Pipeline has run .......................")
   }
 
   def runSingleSampleJobs(sampleConfig: Map[String, Any]): SampleOutput = {
@@ -64,22 +66,36 @@ class Yamsvp(val root: Configurable) extends QScript with MultiSampleQScript {
     var libraryFastqFiles: List[File] = List()
     val sampleID: String = sampleConfig("ID").toString
     val sampleDir: String = outputDir + sampleID + "/"
+    val alignmentDir: String = sampleDir + "alignment/"
+
 
     val svcallingDir: String = sampleDir + "svcalls/"
 
     sampleOutput.libraries = runLibraryJobs(sampleConfig)
     for ((libraryID, libraryOutput) <- sampleOutput.libraries) {
       // this is extending the libraryBamfiles list like '~=' in D or .append in Python or .push_back in C++
-      libraryBamfiles = List(libraryOutput.mappedBamFile)
+      libraryBamfiles ++= List(libraryOutput.mappedBamFile)
     }
-
-    val bamFile: File = if (libraryBamfiles.size == 1) libraryBamfiles.head
-    else if (libraryBamfiles.size > 1) {
-      val mergeSamFiles = MergeSamFiles(this, libraryBamfiles, sampleDir)
-      add(mergeSamFiles)
-      mergeSamFiles.output
-    } else null
-
+    
+    logger.debug( "We have the following bamfiles for merging" )
+    logger.debug( libraryBamfiles )
+
+    val bamFile: File = 
+      if (libraryBamfiles.size == 1) libraryBamfiles.head
+      else if (libraryBamfiles.size > 1) {
+        val mergeSamFiles = new SambambaMerge(root)
+        mergeSamFiles.input = libraryBamfiles
+        mergeSamFiles.output = alignmentDir + sampleID + ".merged.bam"
+        add(mergeSamFiles)
+        
+        val bamIndex = SambambaIndex(root, mergeSamFiles.output)
+        add(bamIndex)
+        
+        mergeSamFiles.output
+      } else null
+
+    logger.debug( bamFile )
+    
     /// bamfile will be used as input for the SV callers. First run Clever
     //    val cleverVCF : File = sampleDir + "/" + sampleID + ".clever.vcf"
 
@@ -88,11 +104,18 @@ class Yamsvp(val root: Configurable) extends QScript with MultiSampleQScript {
     sampleOutput.vcf += ("clever" -> List(clever.outputvcf))
     add(clever)
 
-    val breakdancerDir = sampleDir + sampleID + ".breakdancer/"
+    val breakdancerDir = svcallingDir + sampleID + ".breakdancer/"
     val breakdancer = Breakdancer(this, bamFile, this.reference, breakdancerDir)
     sampleOutput.vcf += ("breakdancer" -> List(breakdancer.outputraw))
     addAll(breakdancer.functions)
-
+//
+//    
+//    val pindelDir = svcallingDir + sampleID + ".pindel/"
+//    val pindel = Pindel(this, bamFile, this.reference, pindelDir)
+//    sampleOutput.vcf += ("pindel" -> List(pindel.outputvcf))
+//    addAll(pindel.functions)
+//
+//    
     return sampleOutput
   }