Merge branch 'feature-FilterOnMappingQualityCarp' into 'develop'

Feature filter on mapping quality carp

Currently testing the pipeline with real samples. Dry run seem to look okay!

Pipeline tested, ready to merge!

See merge request !248

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

+package nl.lumc.sasc.biopet.extensions.picard
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.utils.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{ Output, Input }
+
+/**
+ * Created by sajvanderzeeuw on 6-10-15.
+ */
+class BuildBamIndex(val root: Configurable) extends Picard {
+
+  javaMainClass = new picard.sam.BuildBamIndex().getClass.getName
+
+  @Input(doc = "The input SAM or BAM files to analyze.", required = true)
+  var input: File = _
+
+  @Output(doc = "The output file to bam file to", required = true)
+  var output: File = _
+
+  override def cmdLine = super.cmdLine +
+    required("INPUT=", input, spaceSeparated = false) +
+    required("OUTPUT=", output, spaceSeparated = false)
+}

public/biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/samtools/SamtoolsSort.scala

@@ -3,7 +3,7 @@ package nl.lumc.sasc.biopet.extensions.samtools
 import java.io.File
 
 import nl.lumc.sasc.biopet.utils.config.Configurable
-import org.broadinstitute.gatk.utils.commandline.{Input, Output}
+import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
 
 /**
  * Created by pjvanthof on 22/09/15.

public/biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/samtools/SamtoolsView.scala

@@ -28,7 +28,7 @@ class SamtoolsView(val root: Configurable) extends Samtools {
   @Output(doc = "output File")
   var output: File = null
 
-  var quality: Option[Int] = config("quality")
+  var q: Option[Int] = config("q")
   var b: Boolean = config("b", default = false)
   var h: Boolean = config("h", default = false)
   var f: List[String] = config("f", default = List.empty[String])
@@ -36,7 +36,7 @@ class SamtoolsView(val root: Configurable) extends Samtools {
 
   def cmdBase = required(executable) +
     required("view") +
-    optional("-q", quality) +
+    optional("-q", q) +
     repeat("-f", f) +
     repeat("-F", F) +
     conditional(b, "-b") +

public/carp/src/main/scala/nl/lumc/sasc/biopet/pipelines/carp/Carp.scala

@@ -18,11 +18,12 @@ package nl.lumc.sasc.biopet.pipelines.carp
 import java.io.File
 
 import nl.lumc.sasc.biopet.core._
+import nl.lumc.sasc.biopet.extensions.samtools.SamtoolsView
 import nl.lumc.sasc.biopet.utils.config._
 import nl.lumc.sasc.biopet.core.summary.SummaryQScript
 import nl.lumc.sasc.biopet.extensions.Ln
 import nl.lumc.sasc.biopet.extensions.macs2.Macs2CallPeak
-import nl.lumc.sasc.biopet.extensions.picard.MergeSamFiles
+import nl.lumc.sasc.biopet.extensions.picard.{ BuildBamIndex, MergeSamFiles }
 import nl.lumc.sasc.biopet.pipelines.bammetrics.BamMetrics
 import nl.lumc.sasc.biopet.pipelines.bamtobigwig.Bam2Wig
 import nl.lumc.sasc.biopet.pipelines.mapping.Mapping
@@ -40,8 +41,16 @@ class Carp(val root: Configurable) extends QScript with MultiSampleQScript with
 
   override def defaults = Map(
     "mapping" -> Map(
-      "skip_markduplicates" -> true,
+      "skip_markduplicates" -> false,
       "aligner" -> "bwa-mem"
+    ),
+    "samtoolsview" -> Map("q" -> 10)
+  )
+
+  override def fixedValues = Map(
+    "samtoolsview" -> Map(
+      "h" -> true,
+      "b" -> true
     )
   )
 
@@ -93,6 +102,7 @@ class Carp(val root: Configurable) extends QScript with MultiSampleQScript with
     }
 
     val bamFile = createFile(".bam")
+    val bamFileFilter = createFile(".filter.bam")
     val controls: List[String] = config("control", default = Nil)
 
     def addJobs(): Unit = {
@@ -114,10 +124,29 @@ class Carp(val root: Configurable) extends QScript with MultiSampleQScript with
       val bamMetrics = BamMetrics(qscript, bamFile, new File(sampleDir, "metrics"), sampleId = Some(sampleId))
       addAll(bamMetrics.functions)
       addSummaryQScript(bamMetrics)
+
+      val bamMetricsFilter = BamMetrics(qscript, bamFileFilter, new File(sampleDir, "metrics-filter"), sampleId = Some(sampleId))
+      addAll(bamMetricsFilter.functions)
+      bamMetricsFilter.summaryName = "bammetrics-filter"
+      addSummaryQScript(bamMetricsFilter)
+
       addAll(Bam2Wig(qscript, bamFile).functions)
+      addAll(Bam2Wig(qscript, bamFileFilter).functions)
+
+      val samtoolsView = new SamtoolsView(qscript)
+      samtoolsView.input = bamFile
+      samtoolsView.output = bamFileFilter
+      samtoolsView.b = true
+      samtoolsView.h = true
+      add(samtoolsView)
+
+      val buildBamIndex = new BuildBamIndex(qscript)
+      buildBamIndex.input = bamFileFilter
+      buildBamIndex.output = swapExt(bamFileFilter.getParent, bamFileFilter, ".bam", ".bai")
+      add(buildBamIndex)
 
       val macs2 = new Macs2CallPeak(qscript)
-      macs2.treatment = bamFile
+      macs2.treatment = bamFileFilter
       macs2.name = Some(sampleId)
       macs2.outputdir = sampleDir + File.separator + "macs2" + File.separator + sampleId + File.separator
       add(macs2)
@@ -155,8 +184,8 @@ class Carp(val root: Configurable) extends QScript with MultiSampleQScript with
         if (!samples.contains(controlId))
           throw new IllegalStateException("For sample: " + sampleId + " this control: " + controlId + " does not exist")
         val macs2 = new Macs2CallPeak(this)
-        macs2.treatment = sample.bamFile
-        macs2.control = samples(controlId).bamFile
+        macs2.treatment = sample.bamFileFilter
+        macs2.control = samples(controlId).bamFileFilter
         macs2.name = Some(sampleId + "_VS_" + controlId)
         macs2.outputdir = sample.sampleDir + File.separator + "macs2" + File.separator + macs2.name.get + File.separator
         add(macs2)