Added a general biopet trait for gatk extensions

biopet-framework/examples/gatk-variantcalling.json

@@ -4,5 +4,6 @@
     "haplotypecaller": {
             "stand_call_conf": 20,
             "stand_emit_conf": 20
-        }
+        },
+	"scattercount": 10
 }

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/BiopetJavaCommandLineFunction.scala

@@ -2,7 +2,7 @@ package nl.lumc.sasc.biopet.core
 
 import org.broadinstitute.gatk.queue.function.JavaCommandLineFunction
 
-abstract class BiopetJavaCommandLineFunction extends JavaCommandLineFunction with BiopetCommandLineFunctionTrait {
+trait BiopetJavaCommandLineFunction extends JavaCommandLineFunction with BiopetCommandLineFunctionTrait {
   executable = "java"
 
   override def commandLine: String = {

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

+package nl.lumc.sasc.biopet.extensions.gatk
+
+import nl.lumc.sasc.biopet.core.BiopetJavaCommandLineFunction
+import org.broadinstitute.gatk.queue.extensions.gatk.CommandLineGATK
+
+trait GatkGeneral extends CommandLineGATK with BiopetJavaCommandLineFunction {
+  memoryLimit = Option(2)
+  
+  override val defaultVmem = "4G"
+  
+  if (config.contains("intervals")) intervals = config("intervals").getFileList
+  if (config.contains("exclude_intervals")) excludeIntervals = config("exclude_intervals").getFileList
+  reference_sequence = config("reference")
+  gatk_key = config("gatk_key")
+  if (config.contains("pedigree")) pedigree = config("pedigree").getFileList
+}

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

+package nl.lumc.sasc.biopet.extensions.gatk
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.utils.variant.GATKVCFIndexType
+
+class HaplotypeCaller(val root: Configurable) extends org.broadinstitute.gatk.queue.extensions.gatk.HaplotypeCaller with GatkGeneral {
+  min_mapping_quality_score = config("minMappingQualityScore", default = 20)
+  if (config.contains("scattercount", "haplotypecaller")) scatterCount = config("scattercount")
+  if (config.contains("dbsnp")) this.dbsnp = config("dbsnp")
+  nct = config("threads", default = 3)
+  bamOutput = config("bamOutput")
+  memoryLimit = Option(nct.getOrElse(1) * 2)
+  if (config.contains("allSitePLs")) this.allSitePLs = config("allSitePLs")
+
+  // GVCF options
+  if (config("emitRefConfidence", default = "GVCF").getString == "GVCF") {
+    emitRefConfidence = org.broadinstitute.gatk.tools.walkers.haplotypecaller.ReferenceConfidenceMode.GVCF
+    variant_index_type = GATKVCFIndexType.LINEAR
+    variant_index_parameter = Option(128000)
+  }
+
+  if (config("inputtype", default = "dna").getString == "rna") {
+    dontUseSoftClippedBases = config("dontusesoftclippedbases", default = true)
+    recoverDanglingHeads = config("recoverdanglingheads", default = true)
+    stand_call_conf = config("stand_call_conf", default = 20)
+    stand_emit_conf = config("stand_emit_conf", default = 0)
+  } else {
+    dontUseSoftClippedBases = config("dontusesoftclippedbases", default = false)
+    recoverDanglingHeads = config("recoverdanglingheads", default = false)
+    stand_call_conf = config("stand_call_conf", default = 30)
+    stand_emit_conf = config("stand_emit_conf", default = 0)
+  }
+  
+  override def beforeCmd {
+    super.beforeCmd
+    if (bamOutput != null && nct.getOrElse(1) > 1) {
+      nct = Option(1)
+      logger.warn("BamOutput is on, nct/threads is forced to set on 1, this option is only for debug")
+    }
+  }
+}

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/pipelines/gatk/GatkVariantcalling.scala

@@ -3,8 +3,9 @@ package nl.lumc.sasc.biopet.pipelines.gatk
 import nl.lumc.sasc.biopet.core.{ BiopetQScript, PipelineCommand }
 import nl.lumc.sasc.biopet.core.config.Configurable
 import nl.lumc.sasc.biopet.extensions._
+import nl.lumc.sasc.biopet.extensions.gatk.HaplotypeCaller
 import org.broadinstitute.gatk.queue.QScript
-import org.broadinstitute.gatk.queue.extensions.gatk.{ BaseRecalibrator, CommandLineGATK, HaplotypeCaller, IndelRealigner, PrintReads, RealignerTargetCreator, GenotypeGVCFs, AnalyzeCovariates }
+import org.broadinstitute.gatk.queue.extensions.gatk.{ BaseRecalibrator, CommandLineGATK, IndelRealigner, PrintReads, RealignerTargetCreator, GenotypeGVCFs, AnalyzeCovariates }
 import org.broadinstitute.gatk.queue.function._
 import org.broadinstitute.gatk.utils.commandline.{ Input, Output, Argument }
 import org.broadinstitute.gatk.utils.variant.GATKVCFIndexType
@@ -120,42 +121,9 @@ class GatkVariantcalling(val root: Configurable) extends QScript with BiopetQScr
   }
 
   def addHaplotypeCaller(bamfiles: List[File], outputfile: File): File = {
-    val haplotypeCaller = new HaplotypeCaller with gatkArguments {
-      this.min_mapping_quality_score = config("minMappingQualityScore", 20, "haplotypecaller")
-      if (config.contains("scattercount", "haplotypecaller")) this.scatterCount = config("scattercount", 1, "haplotypecaller")
-      this.input_file = bamfiles
-      this.out = outputfile
-      if (config.contains("dbsnp")) this.dbsnp = config("dbsnp")
-      this.nct = config("threads", default = 3, submodule = "haplotypecaller")
-      if (config("outputToBam", default = false, submodule = "haplotypecaller").getBoolean) {
-        this.bamOutput = outputfile.getAbsolutePath + ".bam"
-        nct = 1
-        logger.warn("BamOutput is on, nct/threads is forced to set on 1, this option is only for debug")
-      }
-      this.memoryLimit = this.nct * 2
-      
-      if (config.contains("allSitePLs")) this.allSitePLs = config("allSitePLs")
-      
-      // GVCF options
-      if (gvcfMode) {
-        this.emitRefConfidence = org.broadinstitute.gatk.tools.walkers.haplotypecaller.ReferenceConfidenceMode.GVCF
-        this.variant_index_type = GATKVCFIndexType.LINEAR
-        this.variant_index_parameter = 128000
-      }
-
-      val inputType: String = config("inputtype", "dna")
-      if (inputType == "rna") {
-        this.dontUseSoftClippedBases = config("dontusesoftclippedbases", true, "haplotypecaller")
-        this.recoverDanglingHeads = config("recoverdanglingheads", true, "haplotypecaller")
-        this.stand_call_conf = config("stand_call_conf", 20, "haplotypecaller")
-        this.stand_emit_conf = config("stand_emit_conf", 20, "haplotypecaller")
-      } else {
-        this.dontUseSoftClippedBases = config("dontusesoftclippedbases", false, "haplotypecaller")
-        this.recoverDanglingHeads = config("recoverdanglingheads", false, "haplotypecaller")
-        this.stand_call_conf = config("stand_call_conf", 30, "haplotypecaller")
-        this.stand_emit_conf = config("stand_emit_conf", 30, "haplotypecaller")
-      }
-    }
+    val haplotypeCaller = new HaplotypeCaller(this)
+    haplotypeCaller.input_file = bamfiles
+    haplotypeCaller.out = outputfile
     add(haplotypeCaller)
 
     return haplotypeCaller.out