Added bam_to_fastq option

biopet-framework/src/main/scala/nl/lumc/sasc/biopet/pipelines/gatk/GatkPipeline.scala

@@ -8,6 +8,7 @@ import scala.collection.JavaConversions._
 import java.io.File
 import nl.lumc.sasc.biopet.extensions.gatk.{ CombineVariants, CombineGVCFs }
 import nl.lumc.sasc.biopet.extensions.picard.AddOrReplaceReadGroups
+import nl.lumc.sasc.biopet.extensions.picard.SamToFastq
 import nl.lumc.sasc.biopet.pipelines.bammetrics.BamMetrics
 import nl.lumc.sasc.biopet.pipelines.mapping.Mapping
 import org.broadinstitute.gatk.queue.QScript
@@ -169,36 +170,48 @@ class GatkPipeline(val root: Configurable) extends QScript with MultiSampleQScri
       var bamFile = new File(runConfig("bam").toString)
       if (!bamFile.exists) throw new IllegalStateException("Bam in config does not exist, file: " + bamFile)
       
-      var readGroupOke = true
-      val inputSam = new SAMFileReader(bamFile)
-      val header = inputSam.getFileHeader.getReadGroups
-      for (readGroup <- inputSam.getFileHeader.getReadGroups) {
-        if (readGroup.getSample != sampleID) logger.warn("Sample ID readgroup in bam file is not the same")
-        if (readGroup.getLibrary != runID) logger.warn("Library ID readgroup in bam file is not the same")
-        if (readGroup.getSample != sampleID || readGroup.getLibrary != runID) readGroupOke = false
-      }
-      inputSam.close
-      
-      if (!readGroupOke) {
-        if (config("correct_readgroups", default = false)) {
-          logger.info("Correcting readgroups, file:" + bamFile)
-          val aorrg = AddOrReplaceReadGroups(this, bamFile, new File(runDir + sampleID + "-" + runID + ".bam"))
-          aorrg.RGID = sampleID + "-" + runID
-          aorrg.RGLB = runID
-          aorrg.RGSM = sampleID
-          if (runConfig.contains("PL")) aorrg.RGPL = runConfig("PL").toString
-          else aorrg.RGPL = "illumina"
-          if (runConfig.contains("PU")) aorrg.RGPU = runConfig("PU").toString
-          else aorrg.RGPU = "na"
-          if (runConfig.contains("CN")) aorrg.RGCN = runConfig("CN").toString
-          add(aorrg, isIntermediate = true)
-          bamFile = aorrg.output
-        } else throw new IllegalStateException("Readgroup sample and/or library of input bamfile is not correct, file: " + bamFile + 
-            "\nPossible to set 'correct_readgroups' to true on config to automatic fix this")
+      if (config("bam_to_fastq", default = false).getBoolean) {
+        val samToFastq = SamToFastq(this, bamFile, runDir + sampleID + "-" + runID + ".R1.fastq", runDir + sampleID + "-" + runID + ".R1.fastq")
+        add(samToFastq)
+        val mapping = Mapping.loadFromLibraryConfig(this, runConfig, sampleConfig, runDir, startJobs = false)
+        mapping.input_R1 = samToFastq.fastqR1
+        mapping.input_R2 = samToFastq.fastqR2
+        mapping.init
+        mapping.biopetScript
+        addAll(mapping.functions) // Add functions of mapping to curent function pool
+        libraryOutput.mappedBamFile = mapping.outputFiles("finalBamFile")
+      } else {
+        var readGroupOke = true
+        val inputSam = new SAMFileReader(bamFile)
+        val header = inputSam.getFileHeader.getReadGroups
+        for (readGroup <- inputSam.getFileHeader.getReadGroups) {
+          if (readGroup.getSample != sampleID) logger.warn("Sample ID readgroup in bam file is not the same")
+          if (readGroup.getLibrary != runID) logger.warn("Library ID readgroup in bam file is not the same")
+          if (readGroup.getSample != sampleID || readGroup.getLibrary != runID) readGroupOke = false
+        }
+        inputSam.close
+
+        if (!readGroupOke) {
+          if (config("correct_readgroups", default = false)) {
+            logger.info("Correcting readgroups, file:" + bamFile)
+            val aorrg = AddOrReplaceReadGroups(this, bamFile, new File(runDir + sampleID + "-" + runID + ".bam"))
+            aorrg.RGID = sampleID + "-" + runID
+            aorrg.RGLB = runID
+            aorrg.RGSM = sampleID
+            if (runConfig.contains("PL")) aorrg.RGPL = runConfig("PL").toString
+            else aorrg.RGPL = "illumina"
+            if (runConfig.contains("PU")) aorrg.RGPU = runConfig("PU").toString
+            else aorrg.RGPU = "na"
+            if (runConfig.contains("CN")) aorrg.RGCN = runConfig("CN").toString
+            add(aorrg, isIntermediate = true)
+            bamFile = aorrg.output
+          } else throw new IllegalStateException("Readgroup sample and/or library of input bamfile is not correct, file: " + bamFile + 
+              "\nPossible to set 'correct_readgroups' to true on config to automatic fix this")
+        }
+        addAll(BamMetrics(this, bamFile, runDir + "metrics/").functions)
+
+        libraryOutput.mappedBamFile = bamFile
       }
-      addAll(BamMetrics(this, bamFile, runDir + "metrics/").functions)
-      
-      libraryOutput.mappedBamFile = bamFile
     } else logger.error("Sample: " + sampleID + ": No R1 found for run: " + runConfig)
     
     val gatkVariantcalling = new GatkVariantcalling(this)

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

@@ -305,7 +305,8 @@ class Mapping(val root: Configurable) extends QScript with BiopetQScript {
 }
 
 object Mapping extends PipelineCommand {
-  def loadFromLibraryConfig(root: Configurable, runConfig: Map[String, Any], sampleConfig: Map[String, Any], runDir: String): Mapping = {
+  def loadFromLibraryConfig(root: Configurable, runConfig: Map[String, Any], sampleConfig: Map[String, Any], 
+                            runDir: String, startJobs: Boolean = true): Mapping = {
     val mapping = new Mapping(root)
 
     logger.debug("Mapping runconfig: " + runConfig)
@@ -322,9 +323,11 @@ object Mapping extends PipelineCommand {
     if (runConfig.contains("PU")) mapping.RGPU = runConfig("PU").toString
     if (runConfig.contains("CN")) mapping.RGCN = runConfig("CN").toString
     mapping.outputDir = runDir
-
-    mapping.init
-    mapping.biopetScript
+    
+    if (startJobs) {
+      mapping.init
+      mapping.biopetScript
+    }
     return mapping
   }
 }