Adding stringtie as expresion measure

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/stringtie/Stringtie.scala

@@ -18,6 +18,9 @@ class Stringtie(val parent: Configurable)
   @Input(required = true)
   var inputBam: File = _
 
+  @Input(required = false)
+  var referenceGtf: Option[File] = None
+
   @Output
   var outputGtf: File = _
 
@@ -52,6 +55,7 @@ class Stringtie(val parent: Configurable)
   def versionRegex: Regex = ".*".r
 
   def cmdLine: String = required(executable) +
+    required(inputBam) +
     conditional(v, "-v") +
     required("-p", threads) +
     conditional(rf, "--rf") +
@@ -70,6 +74,7 @@ class Stringtie(val parent: Configurable)
     optional("-b", b) +
     conditional(e, "-e") +
     optional("-M", M) +
+    optional("-G", referenceGtf) +
     (if (x.nonEmpty) optional("-x", x.mkString(",")) else "") +
     (if (outputAsStdout) "" else required("-o", outputGtf))
 

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/stringtie/StringtieMerge.scala

@@ -16,12 +16,14 @@ class StringtieMerge(val parent: Configurable)
     executable = config("exe", "stringtie")
 
     @Input(required = true)
-    var inputBam: File = _
+    var inputGtfs: List[File] = Nil
+
+    @Input(required = false)
+    var referenceGtf: Option[File] = None
 
     @Output
     var outputGtf: File = _
 
-
     var v: Boolean = config("v", default = logger.isDebugEnabled)
     var l: Option[String] = None
     var f: Option[Double] = config("f")
@@ -38,6 +40,7 @@ class StringtieMerge(val parent: Configurable)
     def versionRegex: Regex = ".*".r
 
     def cmdLine: String = required(executable) +
+      required("--merge")
       conditional(v, "-v") +
       required("-p", threads) +
       optional("-l", l) +
@@ -46,6 +49,8 @@ class StringtieMerge(val parent: Configurable)
       optional("-c", c) +
       optional("-F", F) +
       conditional(i, "-i") +
-      (if (outputAsStdout) "" else required("-o", outputGtf))
+      optional("-G", referenceGtf) +
+      (if (outputAsStdout) "" else required("-o", outputGtf)) +
+      repeat(inputGtfs)
 
   }

gentrap/src/main/scala/nl/lumc/sasc/biopet/pipelines/gentrap/Gentrap.scala

@@ -148,6 +148,7 @@ class Gentrap(val parent: Configurable)
     cufflinksBlind.foreach(validate.gtfFile :+= _.annotationGtf)
     cufflinksGuided.foreach(validate.gtfFile :+= _.annotationGtf)
     cufflinksStrict.foreach(validate.gtfFile :+= _.annotationGtf)
+    stringtie.foreach(validate.gtfFile :+= _.annotationGtf)
     validate.jobOutputFile = new File(outputDir, ".validate.annotation.out")
     add(validate)
 
@@ -169,6 +170,11 @@ class Gentrap(val parent: Configurable)
       Some(new BaseCounts(this))
     else None
 
+  lazy val stringtie: Option[Stringtie] =
+    if (expMeasures().contains(ExpMeasures.Stringtie))
+      Some(new Stringtie(this))
+    else None
+
   lazy val cufflinksBlind: Option[CufflinksBlind] =
     if (expMeasures().contains(ExpMeasures.CufflinksBlind))
       Some(new CufflinksBlind(this))
@@ -186,7 +192,7 @@ class Gentrap(val parent: Configurable)
 
   def executedMeasures: List[QScript with Measurement] =
     (fragmentsPerGene :: baseCounts :: cufflinksBlind ::
-      cufflinksGuided :: cufflinksStrict :: Nil).flatten
+      cufflinksGuided :: cufflinksStrict :: stringtie :: Nil).flatten
 
   /** Whether to do simple variant calling on RNA or not */
   lazy val shivaVariantcalling: Option[ShivaVariantcalling] =
@@ -299,7 +305,7 @@ object Gentrap extends PipelineCommand {
 
   /** Enumeration of available expression measures */
   object ExpMeasures extends Enumeration {
-    val FragmentsPerGene, BaseCounts, CufflinksStrict, CufflinksGuided, CufflinksBlind = Value
+    val FragmentsPerGene, BaseCounts, CufflinksStrict, CufflinksGuided, CufflinksBlind, Stringtie = Value
   }
 
   /** Enumeration of available strandedness */

gentrap/src/main/scala/nl/lumc/sasc/biopet/pipelines/gentrap/measures/Stringtie.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.pipelines.gentrap.measures
+
+import nl.lumc.sasc.biopet.core.annotations.AnnotationGtf
+import nl.lumc.sasc.biopet.extensions.stringtie.{StringtieMerge, Stringtie => StringtieTool}
+import nl.lumc.sasc.biopet.utils.config.Configurable
+import org.broadinstitute.gatk.queue.QScript
+
+/**
+  * Created by pjvan_thof on 1/12/16.
+  */
+class Stringtie(val parent: Configurable)
+    extends QScript
+    with Measurement
+    with AnnotationGtf {
+  def mergeArgs = MergeArgs(idCols = List(1), valCol = 2, fallback = "0")
+
+  /** Pipeline itself */
+  def biopetScript(): Unit = {
+    val sampleGtfFiles: List[File] = bamFiles.map {
+      case (id, file) =>
+        val sampleDir = new File(outputDir, id)
+        val stringtie = new StringtieTool(this)
+        stringtie.inputBam = file
+        stringtie.l = Some(id)
+        stringtie.referenceGtf = Some(annotationGtf)
+        stringtie.outputGtf = new File(sampleDir, s"$id.gtf")
+        stringtie.geneAbundances = Some(new File(sampleDir, s"$id.gene_abund.tab"))
+        stringtie.referenceCoverage = Some(new File(sampleDir, s"$id.cov_refs.gtf"))
+        add(stringtie)
+        stringtie.outputGtf
+    }.toList
+
+    val stringtieMerge = new StringtieMerge(this)
+    stringtieMerge.inputGtfs = sampleGtfFiles
+    stringtieMerge.referenceGtf = Some(annotationGtf)
+    stringtieMerge.outputGtf = new File(outputDir, "stringtie.merged.gtf")
+    add(stringtieMerge)
+
+    addSummaryJobs()
+  }
+
+  override def summaryFiles: Map[String, File] =
+    super.summaryFiles ++ Map("annotation_gtf" -> annotationGtf)
+}