Merge pull request #160 from biopet/fix-BIOPET-747

Upgrade gather methods

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

@@ -17,8 +17,9 @@ 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.scatter.{GATKScatterFunction, LocusScatterFunction}
 import nl.lumc.sasc.biopet.utils.config.Configurable
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class AnalyzeCovariates(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "AnalyzeCovariates"

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Input, Output }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Input, Output}
 
 class ApplyRecalibration(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "ApplyRecalibration"
@@ -41,7 +43,7 @@ class ApplyRecalibration(val parent: Configurable) extends CommandLineGATK with
 
   /** The output filtered and recalibrated VCF file in which each variant is annotated with its VQSLOD value */
   @Output(fullName = "out", shortName = "o", doc = "The output filtered and recalibrated VCF file in which each variant is annotated with its VQSLOD value", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var out: File = _
 
   /** The truth sensitivity level at which to start filtering */

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

@@ -17,10 +17,11 @@ 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.scatter.{ContigScatterFunction, GATKScatterFunction}
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, Input }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Input, Output}
 
 class BaseRecalibrator(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "BaseRecalibrator"

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Gather, Input, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Gather, Input, Output, _}
 
 class CombineGVCFs(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "CombineGVCFs"
@@ -45,7 +47,7 @@ class CombineGVCFs(val parent: Configurable) extends CommandLineGATK with Scatte
 
   /** File to which the combined gVCF should be written */
   @Output(fullName = "out", shortName = "o", doc = "File to which the combined gVCF should be written", required = false, exclusiveOf = "", validation = "")
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var out: File = _
 
   /** If specified, convert banded gVCFs to all-sites gVCFs */

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class CombineVariants(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "CombineVariants"
@@ -33,7 +35,7 @@ class CombineVariants(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 = _
 
   /** Determines how we should merge genotype records for samples shared across the ROD files */

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

@@ -18,11 +18,12 @@ import java.io.File
 
 import nl.lumc.sasc.biopet.core.ScatterGatherableFunction
 import nl.lumc.sasc.biopet.core.summary.Summarizable
+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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
-import org.broadinstitute.gatk.utils.report.{ GATKReport, GATKReportTable }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
+import org.broadinstitute.gatk.utils.report.{GATKReport, GATKReportTable}
 
 class GenotypeConcordance(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction with Summarizable {
   analysisName = "GenotypeConcordance"

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class GenotypeGVCFs(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "GenotypeGVCFs"
@@ -33,7 +35,7 @@ class GenotypeGVCFs(val parent: Configurable) extends CommandLineGATK with Scatt
 
   /** 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 = _
 
   /** Include loci found to be non-variant after genotyping */

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

@@ -17,10 +17,12 @@ 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.{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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Input, Output }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Input, Output}
 
 class HaplotypeCaller(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "HaplotypeCaller"
@@ -29,7 +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 */
@@ -96,7 +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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class IndelRealigner(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "IndelRealigner"
@@ -45,7 +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/MuTect2.scala

@@ -3,6 +3,8 @@ 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.GatherVcfs
+import nl.lumc.sasc.biopet.extensions.gatk.scatter.{GATKScatterFunction, LocusScatterFunction}
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Input, Output}
 
@@ -31,7 +33,7 @@ class MuTect2(val parent: Configurable) extends CommandLineGATK with ScatterGath
 
   /** Output file of the program. */
   @Output(fullName = "out", shortName = "o", required = false)
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var outputVcf: File = _
 
   /**

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

@@ -17,6 +17,8 @@ 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.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._
 
@@ -27,7 +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/RealignerTargetCreator.scala

@@ -17,10 +17,11 @@ 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.scatter.{GATKScatterFunction, LocusScatterFunction}
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Input, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Input, _}
 
 class RealignerTargetCreator(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "RealignerTargetCreator"

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class SelectVariants(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "SelectVariants"
@@ -41,7 +43,7 @@ class SelectVariants(val parent: Configurable) extends CommandLineGATK with Scat
 
   /** 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 = _
 
   /** Include genotypes from this sample */

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Gather, Input, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Gather, Input, Output, _}
 
 class UnifiedGenotyper(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "UnifiedGenotyper"
@@ -221,7 +223,7 @@ class UnifiedGenotyper(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 = _
 
   /** If provided, only these samples will be emitted into the VCF, regardless of which samples are present in the BAM file */

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

@@ -17,10 +17,12 @@ 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.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
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Argument, Gather, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Argument, Gather, Output, _}
 
 class VariantAnnotator(val parent: Configurable) extends CommandLineGATK with ScatterGatherableFunction {
   def analysis_type = "VariantAnnotator"
@@ -49,7 +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/VariantRecalibrator.scala

@@ -16,10 +16,11 @@ package nl.lumc.sasc.biopet.extensions.gatk
 
 import java.io.File
 
+import nl.lumc.sasc.biopet.extensions.gatk.gather.GatherVcfs
 import nl.lumc.sasc.biopet.utils.VcfUtils
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
-import org.broadinstitute.gatk.utils.commandline.{ Gather, Input, Output, _ }
+import org.broadinstitute.gatk.utils.commandline.{Gather, Input, Output, _}
 
 class VariantRecalibrator(val parent: Configurable) extends CommandLineGATK {
   def analysis_type = "VariantRecalibrator"
@@ -122,7 +123,7 @@ class VariantRecalibrator(val parent: Configurable) extends CommandLineGATK {
 
   /** The output recal file used by ApplyRecalibration */
   @Output(fullName = "recal_file", shortName = "recalFile", doc = "The output recal file used by ApplyRecalibration", required = true, exclusiveOf = "", validation = "")
-  @Gather(classOf[CatVariantsGatherer])
+  @Gather(classOf[GatherVcfs])
   var recal_file: File = _
 
   /** The output tranches file used by ApplyRecalibration */

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

@@ -12,14 +12,14 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+package nl.lumc.sasc.biopet.extensions.gatk.gather
 
 import java.io.File
 import java.util
 
 import org.broadinstitute.gatk.engine.recalibration.BQSRGatherer
 import org.broadinstitute.gatk.queue.function.InProcessFunction
-import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
+import org.broadinstitute.gatk.utils.commandline.{Input, Output}
 
 /**
  * Created by pjvanthof on 05/04/2017.

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

@@ -12,9 +12,10 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+package nl.lumc.sasc.biopet.extensions.gatk.gather
 
 import nl.lumc.sasc.biopet.core.BiopetCommandLineFunction
+import nl.lumc.sasc.biopet.extensions.gatk.{CatVariants, CommandLineGATK}
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile
 import org.broadinstitute.gatk.queue.function.scattergather.GatherFunction
 
@@ -26,6 +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

+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 12-7-17.
+  */
+class GatherVcfs extends picard.GatherVcfs(null) with GatherFunction {
+  analysisName = "Gather_Vcfs"
+
+  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/MergeSamFiles.scala → biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/gather/MergeSamFiles.scala

@@ -12,14 +12,17 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+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.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/gatk/ContigScatterFunction.scala → biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/scatter/ContigScatterFunction.scala

@@ -12,7 +12,7 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+package nl.lumc.sasc.biopet.extensions.gatk.scatter
 
 import org.broadinstitute.gatk.queue.function.InProcessFunction
 import org.broadinstitute.gatk.utils.interval.IntervalUtils

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/GATKScatterFunction.scala → biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/scatter/GATKScatterFunction.scala

@@ -12,12 +12,13 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+package nl.lumc.sasc.biopet.extensions.gatk.scatter
 
 import java.io.File
 
+import nl.lumc.sasc.biopet.extensions.gatk.CommandLineGATK
 import org.broadinstitute.gatk.queue.extensions.gatk.GATKIntervals
-import org.broadinstitute.gatk.queue.function.scattergather.{ CloneFunction, ScatterFunction }
+import org.broadinstitute.gatk.queue.function.scattergather.{CloneFunction, ScatterFunction}
 import org.broadinstitute.gatk.utils.commandline.Output
 import org.broadinstitute.gatk.utils.interval.IntervalUtils
 import org.broadinstitute.gatk.utils.io.IOUtils

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/LocusScatterFunction.scala → biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/scatter/LocusScatterFunction.scala

@@ -12,7 +12,7 @@
  * license; For commercial users or users who do not want to follow the AGPL
  * license, please contact us to obtain a separate license.
  */
-package nl.lumc.sasc.biopet.extensions.gatk
+package nl.lumc.sasc.biopet.extensions.gatk.scatter
 
 import org.broadinstitute.gatk.queue.function.InProcessFunction
 import org.broadinstitute.gatk.utils.interval.IntervalUtils

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 = _

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

+/**
+  * Biopet is built on top of GATK Queue for building bioinformatic
+  * pipelines. It is mainly intended to support LUMC SHARK cluster which is running
+  * SGE. But other types of HPC that are supported by GATK Queue (such as PBS)
+  * should also be able to execute Biopet tools and pipelines.
+  *
+  * Copyright 2014 Sequencing Analysis Support Core - Leiden University Medical Center
+  *
+  * Contact us at: sasc@lumc.nl
+  *
+  * A dual licensing mode is applied. The source code within this project is freely available for non-commercial use under an AGPL
+  * license; For commercial users or users who do not want to follow the AGPL
+  * license, please contact us to obtain a separate license.
+  */
+package nl.lumc.sasc.biopet.extensions.picard
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.extensions.Tabix
+import nl.lumc.sasc.biopet.utils.VcfUtils
+import nl.lumc.sasc.biopet.utils.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{Input, Output}
+
+class GatherVcfs(val parent: Configurable) extends Picard {
+
+  @Input(doc = "The input SAM or BAM files to analyze.", required = true)
+  var input: Seq[File] = Nil
+
+  @Output(doc = "The output file to bam file to", required = true)
+  var output: File = _
+
+  //FIXME: This is a workaround for this issue: https://github.com/broadinstitute/picard/issues/789
+  def tabix: Option[Tabix] = if (createIndex) Some(Tabix(this, output)) else None
+
+  @Output(required = false)
+  private var index: File = _
+
+  override def beforeGraph(): Unit = {
+    super.beforeGraph()
+    if (createIndex) index = VcfUtils.getVcfIndexFile(output)
+  }
+
+  override def cmdLine: String =
+    super.cmdLine +
+      repeat("INPUT=", input, spaceSeparated = false) +
+      required("OUTPUT=", output, spaceSeparated = false) + (tabix match {
+      case Some(t) => s" && ${t.cmdLine}"
+      case _ => ""
+    })
+}

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/Shiva.scala

@@ -20,6 +20,7 @@ import nl.lumc.sasc.biopet.core.extensions.CheckValidateVcf
 import nl.lumc.sasc.biopet.core.{PipelineCommand, Reference}
 import nl.lumc.sasc.biopet.core.report.ReportBuilderExtension
 import nl.lumc.sasc.biopet.extensions.gatk._
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.extensions.tools.ValidateVcf
 import nl.lumc.sasc.biopet.pipelines.bammetrics.TargetRegions
 import nl.lumc.sasc.biopet.pipelines.kopisu.Kopisu

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/HaplotypeCaller.scala

@@ -20,7 +20,7 @@
 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
 import nl.lumc.sasc.biopet.extensions.gatk
-import nl.lumc.sasc.biopet.extensions.gatk.BqsrGather
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** Default mode for the haplotypecaller */

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/HaplotypeCallerAllele.scala

@@ -20,7 +20,7 @@
 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
 import nl.lumc.sasc.biopet.extensions.gatk
-import nl.lumc.sasc.biopet.extensions.gatk.BqsrGather
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** Allele mode for Haplotypecaller */

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/UnifiedGenotyper.scala

@@ -20,7 +20,7 @@
 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
 import nl.lumc.sasc.biopet.extensions.gatk
-import nl.lumc.sasc.biopet.extensions.gatk.BqsrGather
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** Default mode for UnifiedGenotyper */

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/UnifiedGenotyperAllele.scala

@@ -20,7 +20,7 @@
 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
 import nl.lumc.sasc.biopet.extensions.gatk
-import nl.lumc.sasc.biopet.extensions.gatk.BqsrGather
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** Allele mode for GenotyperAllele */

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/somatic/MuTect2.scala

 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers.somatic
 
 import nl.lumc.sasc.biopet.extensions.bcftools.BcftoolsReheader
-import nl.lumc.sasc.biopet.extensions.gatk.{BqsrGather, CombineVariants}
+import nl.lumc.sasc.biopet.extensions.gatk.CombineVariants
 import nl.lumc.sasc.biopet.extensions._
+import nl.lumc.sasc.biopet.extensions.gatk.gather.BqsrGather
 import nl.lumc.sasc.biopet.utils.{IoUtils, Logging}
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.queue.extensions.gatk.TaggedFile