Merge branch 'feature-fix_gatk' into 'develop'

Feature fix gatk

Small fix on gatk pipeline

Branch can be removed

See merge request !85

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

@@ -9,40 +9,40 @@ 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 {
-  override def afterGraph {
-    super.afterGraph
-
-    min_mapping_quality_score = config("minMappingQualityScore", default = 20)
-    if (config.contains("scattercount")) scatterCount = config("scattercount")
-    if (config.contains("dbsnp")) this.dbsnp = config("dbsnp")
-    this.sample_ploidy = config("ploidy")
-    nct = config("threads", default = 1)
-    bamOutput = config("bamOutput")
-    memoryLimit = Option(nct.getOrElse(1) * 2)
-    if (config.contains("allSitePLs")) this.allSitePLs = config("allSitePLs")
-    if (config.contains("output_mode")) {
-      import org.broadinstitute.gatk.tools.walkers.genotyper.OutputMode._
-      config("output_mode").asString match {
-        case "EMIT_ALL_CONFIDENT_SITES" => output_mode = EMIT_ALL_CONFIDENT_SITES
-        case "EMIT_ALL_SITES"           => output_mode = EMIT_ALL_SITES
-        case "EMIT_VARIANTS_ONLY"       => output_mode = EMIT_VARIANTS_ONLY
-        case e                          => logger.warn("output mode '" + e + "' does not exist")
-      }
+  min_mapping_quality_score = config("minMappingQualityScore", default = 20)
+  scatterCount = config("scattercount", default = 1)
+  if (config.contains("dbsnp")) this.dbsnp = config("dbsnp")
+  this.sample_ploidy = config("ploidy")
+  if (config.contains("bamOutput")) bamOutput = config("bamOutput")
+  if (config.contains("allSitePLs")) allSitePLs = config("allSitePLs")
+  if (config.contains("output_mode")) {
+    import org.broadinstitute.gatk.tools.walkers.genotyper.OutputMode._
+    config("output_mode").asString match {
+      case "EMIT_ALL_CONFIDENT_SITES" => output_mode = EMIT_ALL_CONFIDENT_SITES
+      case "EMIT_ALL_SITES"           => output_mode = EMIT_ALL_SITES
+      case "EMIT_VARIANTS_ONLY"       => output_mode = EMIT_VARIANTS_ONLY
+      case e                          => logger.warn("output mode '" + e + "' does not exist")
     }
+  }
 
-    if (config("inputtype", default = "dna").asString == "rna") {
-      dontUseSoftClippedBases = config("dontusesoftclippedbases", default = true)
-      stand_call_conf = config("stand_call_conf", default = 5)
-      stand_emit_conf = config("stand_emit_conf", default = 0)
-    } else {
-      dontUseSoftClippedBases = config("dontusesoftclippedbases", default = false)
-      stand_call_conf = config("stand_call_conf", default = 5)
-      stand_emit_conf = config("stand_emit_conf", default = 0)
-    }
+  if (config("inputtype", default = "dna").asString == "rna") {
+    dontUseSoftClippedBases = config("dontusesoftclippedbases", default = true)
+    stand_call_conf = config("stand_call_conf", default = 5)
+    stand_emit_conf = config("stand_emit_conf", default = 0)
+  } else {
+    dontUseSoftClippedBases = config("dontusesoftclippedbases", default = false)
+    stand_call_conf = config("stand_call_conf", default = 5)
+    stand_emit_conf = config("stand_emit_conf", default = 0)
+  }
+
+  override def afterGraph {
+    super.afterGraph
     if (bamOutput != null && nct.getOrElse(1) > 1) {
-      nct = Option(1)
+      threads = 1
       logger.warn("BamOutput is on, nct/threads is forced to set on 1, this option is only for debug")
     }
+    nct = Some(threads)
+    memoryLimit = Option(memoryLimit.getOrElse(2.0) * nct.getOrElse(1))
   }
 
   def useGvcf() {

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

@@ -50,7 +50,7 @@ class GatkVariantcalling(val root: Configurable) extends QScript with BiopetQScr
 
   def init() {
     if (outputName == null && sampleID != null) outputName = sampleID
-    else if (outputName == null) outputName = "noname"
+    else if (outputName == null) outputName = config("output_name", default = "noname")
     if (outputDir == null) throw new IllegalStateException("Missing Output directory on gatk module")
     else if (!outputDir.endsWith("/")) outputDir += "/"
 

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

@@ -15,13 +15,11 @@
  */
 package nl.lumc.sasc.biopet.core
 
-//import java.io.BufferedInputStream
 import java.io.File
 import nl.lumc.sasc.biopet.core.config.Configurable
 import org.broadinstitute.gatk.queue.QException
 import org.broadinstitute.gatk.queue.function.CommandLineFunction
 import org.broadinstitute.gatk.utils.commandline.{ Input, Argument }
-//import scala.io.Source
 import scala.sys.process.{ Process, ProcessLogger }
 import scala.util.matching.Regex
 import java.io.FileInputStream
@@ -53,7 +51,7 @@ trait BiopetCommandLineFunctionTrait extends CommandLineFunction with Configurab
   override def freezeFieldValues() {
     checkExecutable
     afterGraph
-    jobOutputFile = new File(firstOutput.getParent + "/." + firstOutput.getName + "." + configName + ".out")
+    if (jobOutputFile == null) jobOutputFile = new File(firstOutput.getParent + "/." + firstOutput.getName + "." + configName + ".out")
 
     if (threads == 0) threads = getThreads(defaultThreads)
     if (threads > 1) nCoresRequest = Option(threads)
@@ -62,8 +60,8 @@ trait BiopetCommandLineFunctionTrait extends CommandLineFunction with Configurab
       vmem = config("vmem")
       if (vmem.isEmpty && defaultVmem.nonEmpty) vmem = Some(defaultVmem)
     }
-    if (vmem != null) jobResourceRequests :+= "h_vmem=" + vmem
-    jobName = configName + ":" + firstOutput.getName
+    if (vmem.isDefined) jobResourceRequests :+= "h_vmem=" + vmem.get
+    jobName = configName + ":" + (if (firstOutput != null) firstOutput.getName else jobOutputFile)
 
     super.freezeFieldValues()
   }