Updated krakentojson report and enforce check-names with-in Kraken

public/biopet-extentsions/src/main/scala/nl/lumc/sasc/biopet/extensions/kraken/Kraken.scala

@@ -39,11 +39,6 @@ class Kraken(val root: Configurable) extends BiopetCommandLineFunction {
 
   var db: File = config("db")
 
-  var inputFastQ: Boolean = true
-  var compression: Boolean = false
-  var compressionGzip: Boolean = false
-  var compressionBzip: Boolean = false
-
   var quick: Boolean = false
   var minHits: Option[Int] = config("min_hits")
 
@@ -68,8 +63,6 @@ class Kraken(val root: Configurable) extends BiopetCommandLineFunction {
   def cmdLine = required(executable) +
     "--db" + required(db) +
     optional("--threads", nCoresRequest) +
-    conditional(inputFastQ, "--fastq-input") +
-    conditional(!inputFastQ, "--fasta-input") +
     conditional(quick, "--quick") +
     optional("--min_hits", minHits) +
     optional("--unclassified-out ", unclassified_out.get) +
@@ -77,5 +70,6 @@ class Kraken(val root: Configurable) extends BiopetCommandLineFunction {
     "--output" + required(output) +
     conditional(preLoad, "--preload") +
     conditional(paired, "--paired") +
+    conditional(paired, "--check-names") +
     repeat(input)
 }

public/biopet-extentsions/src/main/scala/nl/lumc/sasc/biopet/extensions/kraken/KrakenReport.scala

@@ -44,9 +44,9 @@ class KrakenReport(val root: Configurable) extends BiopetCommandLineFunction {
   var output: File = _
 
   def cmdLine: String = {
-    val cmd: String = "--db " + required(db) +
+    val cmd: String = required(executable) + "--db " + required(db) +
       conditional(show_zeros, "--show-zeros") +
-      input.getAbsolutePath + ">" + output.getAbsolutePath
+      required(input.getAbsolutePath) + " > " + required(output.getAbsolutePath)
     cmd
   }
 }

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

@@ -24,14 +24,17 @@ class KrakenReportToJson(val root: Configurable) extends ToolCommandFuntion with
   var inputReport: File = null
 
   @Argument(required = false)
-  var skipNames: Boolean = true
+  var skipNames: Boolean = false
 
   @Output(doc = "Output JSON", shortName = "output", required = true)
   var output: File = null
 
   override def defaultCoreMemory = 1.0
 
-  override def cmdLine = super.cmdLine + required("-i", inputReport) + required("-o", output)
+  override def cmdLine = super.cmdLine +
+                      required("-i", inputReport) +
+                      required("-o", output) +
+                      conditional(skipNames, "--skipnames")
 
   def summaryStats: Map[String, Any] = {
     val map = ConfigUtils.fileToConfigMap(output)

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

@@ -56,7 +56,7 @@ object KrakenReportToJson extends ToolCommand {
   var cladeIDs: mutable.ArrayBuffer[Long] = mutable.ArrayBuffer.fill(32)(0)
   val spacePattern = "^( +)".r
 
-  case class Args(krakenreport: File = null, outputJson: Option[File] = None, skipNames: Boolean = true) extends AbstractArgs
+  case class Args(krakenreport: File = null, outputJson: Option[File] = None, skipNames: Boolean = false) extends AbstractArgs
 
   class OptParser extends AbstractOptParser {
 
@@ -93,6 +93,9 @@ object KrakenReportToJson extends ToolCommand {
 
   def parseLine(krakenRawHit: String, skipNames: Boolean): Map[Long, KrakenHit] = {
     val values: Array[String] = krakenRawHit.stripLineEnd.split("\t")
+
+    assert(values.length == 6)
+
     val scientificName: String = values(5)
     val cladeLevel = spacePattern.findFirstIn(scientificName).getOrElse("").length / 2
 

public/gears/src/main/scala/nl/lumc/sasc/biopet/pipelines/gears/Gears.scala

@@ -74,16 +74,10 @@ class Gears(val root: Configurable) extends QScript with SummaryQScript {
       samFilterUnmapped.isIntermediate = false
       add(samFilterUnmapped)
 
-      val samNameSort = new SortSam(qscript)
-      samNameSort.input = samFilterUnmapped.output
-      samNameSort.output = new File(outputDir, s"$outputName.unmapped.nsort.bam")
-      samNameSort.sortOrder = "queryname"
-      samNameSort.isIntermediate = false
-      add(samNameSort)
-
       // start bam to fastq (only on unaligned reads) also extract the matesam
       val samToFastq = new SamToFastq(qscript)
-      samToFastq.input= samNameSort.output
+      samToFastq.input= samFilterUnmapped.output
+      samToFastq.stringency = Some("LENIENT")
       samToFastq.fastqR1 = new File(outputDir, s"$outputName.unmapped.R1.fq.gz")
       samToFastq.fastqR2 = new File(outputDir, s"$outputName.unmapped.R2.fq.gz")
       samToFastq.fastqUnpaired = new File(outputDir, s"$outputName.unmapped.singleton.fq.gz")
@@ -136,9 +130,9 @@ class Gears(val root: Configurable) extends QScript with SummaryQScript {
     GearsOutputFiles ++ Map("kraken_report_output" -> krakenReport.output)
 
     val krakenReportJSON = new KrakenReportToJson(qscript)
-    krakenReportJSON.inputReport = krakenAnalysis.output
+    krakenReportJSON.inputReport = krakenReport.output
     krakenReportJSON.output = new File(outputDir, s"$outputName.krkn.json")
-    krakenReportJSON.skipNames = config("skipNames", default = true)
+    krakenReportJSON.skipNames = config("skipNames", default = false)
     add(krakenReportJSON)
 
     addSummaryJobs()