Merge branch 'patch-mapping' into 'develop'

Patch mapping

Fix for #122

See merge request !120

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

@@ -20,7 +20,7 @@ import java.io.File
 import java.util.Date
 import nl.lumc.sasc.biopet.core.summary.SummaryQScript
 import nl.lumc.sasc.biopet.core.{ SampleLibraryTag, BiopetQScript, PipelineCommand }
-import nl.lumc.sasc.biopet.extensions.{ Ln, Star, Stampy, Bowtie }
+import nl.lumc.sasc.biopet.extensions._
 import nl.lumc.sasc.biopet.extensions.bwa.{ BwaSamse, BwaSampe, BwaAln, BwaMem }
 import nl.lumc.sasc.biopet.pipelines.bamtobigwig.Bam2Wig
 import nl.lumc.sasc.biopet.tools.FastqSplitter
@@ -163,9 +163,14 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
       chunks += (chunkDir -> (removeGz(chunkDir + input_R1.getName),
         if (paired) removeGz(chunkDir + input_R2.get.getName) else ""))
     }
-    else chunks += (outputDir -> (
-      flexiprep.extractIfNeeded(input_R1, flexiprep.outputDir),
-      if (paired) flexiprep.extractIfNeeded(input_R2.get, flexiprep.outputDir) else "")
+    else if (skipFlexiprep) {
+      chunks += (outputDir -> (
+        extractIfNeeded(input_R1, flexiprep.outputDir),
+        if (paired) extractIfNeeded(input_R2.get, outputDir) else "")
+      )
+    } else chunks += (outputDir -> (
+      flexiprep.outputFiles("fastq_input_R1"),
+      if (paired) flexiprep.outputFiles("fastq_input_R2") else "")
     )
 
     if (chunking) {
@@ -441,6 +446,32 @@ class Mapping(val root: Configurable) extends QScript with SummaryQScript with S
 
     return RG.substring(0, RG.lastIndexOf("\\t"))
   }
+
+  //FIXME: This is code duplication from flexiprep, need general class to pass jobs inside a util function
+  /**
+   * Extracts file if file is compressed
+   * @param file
+   * @param runDir
+   * @return returns extracted file
+   */
+  def extractIfNeeded(file: File, runDir: File): File = {
+    if (file == null) return file
+    else if (file.getName().endsWith(".gz") || file.getName().endsWith(".gzip")) {
+      var newFile: File = swapExt(runDir, file, ".gz", "")
+      if (file.getName().endsWith(".gzip")) newFile = swapExt(runDir, file, ".gzip", "")
+      val zcatCommand = Zcat(this, file, newFile)
+      zcatCommand.isIntermediate = true
+      add(zcatCommand)
+      return newFile
+    } else if (file.getName().endsWith(".bz2")) {
+      val newFile = swapExt(runDir, file, ".bz2", "")
+      val pbzip2 = Pbzip2(this, file, newFile)
+      pbzip2.isIntermediate = true
+      add(pbzip2)
+      return newFile
+    } else return file
+  }
+
 }
 
 object Mapping extends PipelineCommand
\ No newline at end of file

public/mapping/src/test/scala/nl/lumc/sasc/biopet/pipelines/mapping/MappingTest.scala

@@ -37,18 +37,23 @@ class MappingTest extends TestNGSuite with Matchers {
     val chunks = Array(1, 5, 10, 100)
     val skipMarkDuplicates = Array(true, false)
     val skipFlexipreps = Array(true, false)
+    val zipped = Array(true, false)
 
     for (
       aligner <- aligners;
       pair <- paired;
       chunk <- chunks;
       skipMarkDuplicate <- skipMarkDuplicates;
-      skipFlexiprep <- skipFlexipreps
-    ) yield Array(aligner, pair, chunk, skipMarkDuplicate, skipFlexiprep)
+      skipFlexiprep <- skipFlexipreps;
+      zipped <- zipped
+    ) yield Array(aligner, pair, chunk, skipMarkDuplicate, skipFlexiprep, zipped)
   }
 
   @Test(dataProvider = "mappingOptions")
-  def testMapping(aligner: String, paired: Boolean, chunks: Int, skipMarkDuplicate: Boolean, skipFlexiprep: Boolean) = {
+  def testMapping(aligner: String, paired: Boolean, chunks: Int,
+                  skipMarkDuplicate: Boolean,
+                  skipFlexiprep: Boolean,
+                  zipped: Boolean) = {
     val map = ConfigUtils.mergeMaps(Map("output_dir" -> MappingTest.outputDir,
       "aligner" -> aligner,
       "number_chunks" -> chunks,
@@ -57,15 +62,20 @@ class MappingTest extends TestNGSuite with Matchers {
     ), Map(MappingTest.executables.toSeq: _*))
     val mapping: Mapping = initPipeline(map)
 
-    mapping.input_R1 = new File(mapping.outputDir, "bla_R1.fq")
-    if (paired) mapping.input_R2 = Some(new File(mapping.outputDir, "bla_R2.fq"))
+    if (zipped) {
+      mapping.input_R1 = new File(mapping.outputDir, "bla_R1.fq.gz")
+      if (paired) mapping.input_R2 = Some(new File(mapping.outputDir, "bla_R2.fq.gz"))
+    } else {
+      mapping.input_R1 = new File(mapping.outputDir, "bla_R1.fq")
+      if (paired) mapping.input_R2 = Some(new File(mapping.outputDir, "bla_R2.fq"))
+    }
     mapping.sampleId = Some("1")
     mapping.libId = Some("1")
     mapping.script()
 
     //Flexiprep
     mapping.functions.count(_.isInstanceOf[Fastqc]) shouldBe (if (skipFlexiprep) 0 else if (paired) 4 else 2)
-    mapping.functions.count(_.isInstanceOf[Zcat]) shouldBe 0
+    mapping.functions.count(_.isInstanceOf[Zcat]) shouldBe (if (!zipped || (chunks > 1 && skipFlexiprep)) 0 else if (paired) 2 else 1)
     mapping.functions.count(_.isInstanceOf[Seqstat]) shouldBe ((if (skipFlexiprep) 0 else if (paired) 4 else 2) * chunks)
     mapping.functions.count(_.isInstanceOf[SeqtkSeq]) shouldBe ((if (skipFlexiprep) 0 else if (paired) 2 else 1) * chunks)
     mapping.functions.count(_.isInstanceOf[Cutadapt]) shouldBe ((if (skipFlexiprep) 0 else if (paired) 2 else 1) * chunks)