Switch to GatherBamFiles

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/HaplotypeCaller.scala

@@ -17,7 +17,7 @@ package nl.lumc.sasc.biopet.extensions.gatk
 import java.io.File
 
 import nl.lumc.sasc.biopet.core.ScatterGatherableFunction
-import nl.lumc.sasc.biopet.extensions.gatk.gather.{CatVariantsGatherer, MergeSamFiles}
+import nl.lumc.sasc.biopet.extensions.gatk.gather.{GatherVcfs, GatherBamFiles}
 import nl.lumc.sasc.biopet.extensions.gatk.scatter.{GATKScatterFunction, LocusScatterFunction}
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
@@ -31,7 +31,7 @@ class HaplotypeCaller(val parent: Configurable) extends CommandLineGATK with Sca
 
   /** File to which variants should be written */
   @Output(fullName = "out", shortName = "o", doc = "File to which variants should be written", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var out: File = _
 
   /** What likelihood calculation engine to use to calculate the relative likelihood of reads vs haplotypes */
@@ -98,7 +98,7 @@ class HaplotypeCaller(val parent: Configurable) extends CommandLineGATK with Sca
 
   /** File to which assembled haplotypes should be written */
   @Output(fullName = "bamOutput", shortName = "bamout", doc = "File to which assembled haplotypes should be written", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[MergeSamFiles])
+  @Gather(classOf[GatherBamFiles])
   var bamOutput: File = _
 
   /** Automatically generated md5 for bamOutput */

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/IndelRealigner.scala

@@ -17,7 +17,7 @@ package nl.lumc.sasc.biopet.extensions.gatk
 import java.io.File
 
 import nl.lumc.sasc.biopet.core.ScatterGatherableFunction
-import nl.lumc.sasc.biopet.extensions.gatk.gather.MergeSamFiles
+import nl.lumc.sasc.biopet.extensions.gatk.gather.GatherBamFiles
 import nl.lumc.sasc.biopet.extensions.gatk.scatter.{ContigScatterFunction, GATKScatterFunction}
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
@@ -47,7 +47,7 @@ class IndelRealigner(val parent: Configurable) extends CommandLineGATK with Scat
 
   /** Output bam */
   @Output(fullName = "out", shortName = "o", doc = "Output bam", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[MergeSamFiles])
+  @Gather(classOf[GatherBamFiles])
   var out: File = _
 
   /** Automatically generated md5 for out */

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/PrintReads.scala

@@ -17,7 +17,7 @@ package nl.lumc.sasc.biopet.extensions.gatk
 import java.io.File
 
 import nl.lumc.sasc.biopet.core.ScatterGatherableFunction
-import nl.lumc.sasc.biopet.extensions.gatk.gather.MergeSamFiles
+import nl.lumc.sasc.biopet.extensions.gatk.gather.GatherBamFiles
 import nl.lumc.sasc.biopet.extensions.gatk.scatter.{ContigScatterFunction, GATKScatterFunction}
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.utils.commandline._
@@ -29,7 +29,7 @@ class PrintReads(val parent: Configurable) extends CommandLineGATK with ScatterG
 
   /** Write output to this BAM filename instead of STDOUT */
   @Output(fullName = "out", shortName = "o", doc = "Write output to this BAM filename instead of STDOUT", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[MergeSamFiles])
+  @Gather(classOf[GatherBamFiles])
   var out: File = _
 
   /** Exclude all reads with this read group from the output */

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/VariantAnnotator.scala

@@ -17,7 +17,7 @@ package nl.lumc.sasc.biopet.extensions.gatk
 import java.io.File
 
 import nl.lumc.sasc.biopet.core.ScatterGatherableFunction
-import nl.lumc.sasc.biopet.extensions.gatk.gather.CatVariantsGatherer
+import nl.lumc.sasc.biopet.extensions.gatk.gather.GatherVcfs
 import nl.lumc.sasc.biopet.extensions.gatk.scatter.{GATKScatterFunction, LocusScatterFunction}
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
@@ -51,7 +51,7 @@ class VariantAnnotator(val parent: Configurable) extends CommandLineGATK with Sc
 
   /** File to which variants should be written */
   @Output(fullName = "out", shortName = "o", doc = "File to which variants should be written", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var out: File = _
 
   /** One or more specific annotations to apply to variant calls */

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/gather/CatVariantsGatherer.scala

@@ -27,6 +27,8 @@ import org.broadinstitute.gatk.queue.function.scattergather.GatherFunction
  *           \@Gather(className = "org.broadinstitute.gatk.queue.extensions.gatk.CatVariantsGatherer")
  *           \@Output(doc="File to which variants should be written",required=true)
  *           protected VariantContextWriter writer = null;
+ *
+ *           @deprecated
  */
 class CatVariantsGatherer extends CatVariants(null) with GatherFunction {
   this.assumeSorted = true

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/gather/GatherBamFiles.scala

+package nl.lumc.sasc.biopet.extensions.gatk.gather
+
+import nl.lumc.sasc.biopet.core.BiopetCommandLineFunction
+import nl.lumc.sasc.biopet.extensions.gatk.CommandLineGATK
+import nl.lumc.sasc.biopet.extensions.picard
+import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
+import org.broadinstitute.gatk.queue.function.scattergather.GatherFunction
+
+/**
+  * Created by pjvan_thof on 13-7-17.
+  */
+class GatherBamFiles extends picard.GatherBamFiles(null) with GatherFunction {
+  analysisName = "Gather_Bam_Files"
+
+  override val parent = originalFunction match {
+    case b: BiopetCommandLineFunction => b
+    case _                            => null
+  }
+
+  override def freezeFieldValues() {
+    val originalGATK = this.originalFunction.asInstanceOf[CommandLineGATK]
+
+    this.input = this.gatherParts.zipWithIndex map { case (input, index) => new TaggedFile(input, "input" + index) }
+    this.output = this.originalOutput
+
+    super.freezeFieldValues()
+  }
+
+}

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/gather/GatherVcfs.scala

@@ -10,7 +10,7 @@ import org.broadinstitute.gatk.queue.function.scattergather.GatherFunction
   * Created by pjvan_thof on 12-7-17.
   */
 class GatherVcfs extends picard.GatherVcfs(null) with GatherFunction {
-  analysisName = "Gather_CatVariants"
+  analysisName = "Gather_Vcfs"
 
   override val parent = originalFunction match {
     case b: BiopetCommandLineFunction => b

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/gather/MergeSamFiles.scala

@@ -21,6 +21,8 @@ import org.broadinstitute.gatk.queue.function.scattergather.GatherFunction
 
 /**
  * Merges BAM files using htsjdk.samtools.MergeSamFiles.
+ *
+ * @deprecated
  */
 class MergeSamFiles extends picard.MergeSamFiles(null) with GatherFunction {
 

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/picard/GatherBamFiles.scala

@@ -22,7 +22,7 @@ import org.broadinstitute.gatk.utils.commandline.{Input, Output}
 class GatherBamFiles(val parent: Configurable) extends Picard {
 
   @Input(doc = "The input SAM or BAM files to analyze.", required = true)
-  var input: List[File] = Nil
+  var input: Seq[File] = Nil
 
   @Output(doc = "The output file to bam file to", required = true)
   var output: File = _