Merge branch 'fix_gsnap_piping' into 'develop'

Changes in gsnap piping and fix bug in fastqsync

This is I think a forgotten branch.

@s.a.j.van_der_zeeuw: Is this tested and does this work?

See merge request !270

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

@@ -331,6 +331,14 @@ class Gsnap(val root: Configurable) extends BiopetCommandLineFunction with Refer
   def versionRegex = """.* version (.*)""".r
   def versionCommand = executable + " --version"
 
+  override def beforeGraph(): Unit = {
+    super.beforeGraph()
+    if ((!gunzip && !bunzip2) && input.forall(_.getName.endsWith(".gz"))) {
+      logger.info("Fastq with .gz extension found, enabled --gunzip option")
+      gunzip = true
+    }
+  }
+
   def cmdLine = {
     required(executable) +
       optional("--dir", dir) +

public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/FastqSync.scala

@@ -26,11 +26,11 @@ import scala.collection.JavaConverters._
 object FastqSync extends ToolCommand {
 
   /** Regex for capturing read ID ~ taking into account its read pair mark (if present) */
-  private val idRegex = "[_/][12]\\s??|\\s".r
+  private val idRegex = """[_/][12]$""".r
 
   /** Implicit class to allow for lazy retrieval of FastqRecord ID without any read pair mark */
   private implicit class FastqPair(fq: FastqRecord) {
-    lazy val fragId = idRegex.split(fq.getReadHeader)(0)
+    lazy val fragId = idRegex.split(fq.getReadHeader.split(" ")(0))(0)
   }
 
   /**

public/biopet-tools/src/test/scala/nl/lumc/sasc/biopet/tools/FastqSyncTest.scala

@@ -188,6 +188,32 @@ class FastqSyncTest extends TestNGSuite with MockitoSugar with Matchers {
     numKept shouldBe 1
   }
 
+  @Test(dataProvider = "mockProvider")
+  def testSeqSolexa(refMock: FastqReader, aMock: FastqReader, bMock: FastqReader,
+                    aOutMock: AsyncFastqWriter, bOutMock: AsyncFastqWriter) = {
+
+    when(refMock.iterator) thenReturn recordsOver(
+      "SOLEXA12_24:6:117:1388:2001/2",
+      "SOLEXA12_24:6:96:470:1965/2",
+      "SOLEXA12_24:6:35:1209:2037/2")
+    when(aMock.iterator) thenReturn recordsOver(
+      "SOLEXA12_24:6:96:470:1965/1",
+      "SOLEXA12_24:6:35:1209:2037/1")
+    when(bMock.iterator) thenReturn recordsOver(
+      "SOLEXA12_24:6:117:1388:2001/2",
+      "SOLEXA12_24:6:96:470:1965/2")
+    val obs = inOrd(aOutMock, bOutMock)
+
+    val (numDiscard1, numDiscard2, numKept) = syncFastq(refMock, aMock, bMock, aOutMock, bOutMock)
+
+    obs.verify(aOutMock).write(new FastqRecord("SOLEXA12_24:6:96:470:1965/1", "A", "", "H"))
+    obs.verify(bOutMock).write(new FastqRecord("SOLEXA12_24:6:96:470:1965/2", "A", "", "H"))
+
+    numDiscard1 shouldBe 1
+    numDiscard2 shouldBe 1
+    numKept shouldBe 1
+  }
+
   @Test(dataProvider = "mockProvider")
   def testSeqABShorterPairMarkSlash(refMock: FastqReader, aMock: FastqReader, bMock: FastqReader,
                                     aOutMock: AsyncFastqWriter, bOutMock: AsyncFastqWriter) = {

public/mapping/src/main/scala/nl/lumc/sasc/biopet/pipelines/mapping/Mapping.scala

@@ -334,10 +334,8 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
   }
 
   def addGsnap(R1: File, R2: Option[File], output: File): File = {
-    val zcatR1 = extractIfNeeded(R1, output.getParentFile)
-    val zcatR2 = if (paired) Some(extractIfNeeded(R2.get, output.getParentFile)) else None
     val gsnapCommand = new Gsnap(this)
-    gsnapCommand.input = if (paired) List(zcatR1._2, zcatR2.get._2) else List(zcatR1._2)
+    gsnapCommand.input = if (paired) List(R1, R2.get) else List(R1)
     gsnapCommand.output = swapExt(output.getParentFile, output, ".bam", ".sam")
 
     val reorderSam = new ReorderSam(this)
@@ -345,11 +343,8 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
     reorderSam.output = swapExt(output.getParentFile, output, ".sorted.bam", ".reordered.bam")
 
     val ar = addAddOrReplaceReadGroups(reorderSam.output, output)
-    val pipe = new BiopetFifoPipe(this, (zcatR1._1 :: (if (paired) zcatR2.get._1 else None) ::
-      Some(gsnapCommand) :: Some(ar._1) :: Some(reorderSam) :: Nil).flatten)
-    pipe.threadsCorrection = -1
-    zcatR1._1.foreach(x => pipe.threadsCorrection -= 1)
-    zcatR2.foreach(_._1.foreach(x => pipe.threadsCorrection -= 1))
+    val pipe = new BiopetFifoPipe(this, gsnapCommand :: ar._1 :: reorderSam :: Nil)
+    pipe.threadsCorrection = -2
     add(pipe)
     ar._2
   }