Added summary to VcfStats

protected/biopet-gatk-pipelines/src/main/scala/nl/lumc/sasc/biopet/pipelines/gatk/ShivaVariantcallingGatk.scala

@@ -27,7 +27,7 @@ class ShivaVariantcallingGatk(val root: Configurable) extends QScript with Shiva
     protected val defaultPrio = 1
     protected val defaultUse = true
 
-    def outputFile = new File(outputDir, namePrefix + "haplotypecaller.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".haplotypecaller.vcf.gz")
 
     def addJobs() {
       val hc = new nl.lumc.sasc.biopet.extensions.gatk.broad.HaplotypeCaller(qscript)
@@ -42,7 +42,7 @@ class ShivaVariantcallingGatk(val root: Configurable) extends QScript with Shiva
     protected val defaultPrio = 20
     protected val defaultUse = false
 
-    def outputFile = new File(outputDir, namePrefix + "unifiedgenotyper.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".unifiedgenotyper.vcf.gz")
 
     def addJobs() {
       val ug = new nl.lumc.sasc.biopet.extensions.gatk.broad.UnifiedGenotyper(qscript)
@@ -57,7 +57,7 @@ class ShivaVariantcallingGatk(val root: Configurable) extends QScript with Shiva
     protected val defaultPrio = 5
     protected val defaultUse = false
 
-    def outputFile = new File(outputDir, namePrefix + "haplotypecaller_allele.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".haplotypecaller_allele.vcf.gz")
 
     def addJobs() {
       val hc = new nl.lumc.sasc.biopet.extensions.gatk.broad.HaplotypeCaller(qscript)
@@ -74,7 +74,7 @@ class ShivaVariantcallingGatk(val root: Configurable) extends QScript with Shiva
     protected val defaultPrio = 9
     protected val defaultUse = false
 
-    def outputFile = new File(outputDir, namePrefix + "unifiedgenotyper_allele.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".unifiedgenotyper_allele.vcf.gz")
 
     def addJobs() {
       val ug = new nl.lumc.sasc.biopet.extensions.gatk.broad.UnifiedGenotyper(qscript)
@@ -91,7 +91,7 @@ class ShivaVariantcallingGatk(val root: Configurable) extends QScript with Shiva
     protected val defaultPrio = 5
     protected val defaultUse = false
 
-    def outputFile = new File(outputDir, namePrefix + "haplotypecaller_gvcf.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".haplotypecaller_gvcf.vcf.gz")
 
     def addJobs() {
       val gvcfFiles = for (inputBam <- inputBams) yield {

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

@@ -4,23 +4,31 @@ import java.io.{ FileOutputStream, PrintWriter, File }
 
 import htsjdk.variant.variantcontext.{ VariantContext, Genotype }
 import htsjdk.variant.vcf.VCFFileReader
+import nl.lumc.sasc.biopet.core.summary.Summarizable
 import nl.lumc.sasc.biopet.core.{ BiopetJavaCommandLineFunction, ToolCommand }
 import nl.lumc.sasc.biopet.core.config.Configurable
 import org.broadinstitute.gatk.utils.commandline.{ Output, Input }
 import scala.collection.JavaConversions._
 import scala.collection.mutable
+import scala.io.Source
 import scala.sys.process.{ Process, ProcessLogger }
 import htsjdk.samtools.util.Interval
 
 /**
  * Created by pjvan_thof on 1/10/15.
  */
-class VcfStats(val root: Configurable) extends BiopetJavaCommandLineFunction {
+class VcfStats(val root: Configurable) extends BiopetJavaCommandLineFunction with Summarizable {
   javaMainClass = getClass.getName
 
   @Input(doc = "Input fastq", shortName = "I", required = true)
   var input: File = _
 
+  @Output
+  protected var generalStats: File = null
+
+  @Output
+  protected var genotypeStats: File = null
+
   override val defaultVmem = "4G"
   override val defaultThreads = 3
 
@@ -32,7 +40,9 @@ class VcfStats(val root: Configurable) extends BiopetJavaCommandLineFunction {
    */
   def setOutputDir(dir: File): Unit = {
     outputDir = dir
-    this.jobOutputFile = new File(dir, ".vcfstats.out")
+    generalStats = new File(dir, "general.tsv")
+    genotypeStats = new File(dir, "genotype_general.tsv")
+    jobOutputFile = new File(dir, ".vcfstats.out")
   }
 
   /**
@@ -42,6 +52,27 @@ class VcfStats(val root: Configurable) extends BiopetJavaCommandLineFunction {
   override def commandLine = super.commandLine +
     required("-I", input) +
     required("-o", outputDir)
+
+  /**
+   * Returns general stats to the summary
+   * @return
+   */
+  def summaryStats: Map[String, Any] = {
+    (for (
+      line <- Source.fromFile(generalStats).getLines();
+      values = line.split("\t") if values.size >= 2 && !values(0).isEmpty
+    ) yield values(0) -> values(1).toInt
+    ).toMap
+  }
+
+  /**
+   * return only general files to summary
+   * @return
+   */
+  def summaryFiles: Map[String, File] = Map(
+    "general_stats" -> generalStats,
+    "genotype_stats" -> genotypeStats
+  )
 }
 
 object VcfStats extends ToolCommand {

public/shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/ShivaVariantcallingTrait.scala

@@ -22,16 +22,16 @@ trait ShivaVariantcallingTrait extends SummaryQScript with SampleLibraryTag {
 
   def namePrefix: String = {
     (sampleId, libId) match {
-      case (Some(sampleId), Some(libId)) => sampleId + "-" + libId + "."
-      case (Some(sampleId), _)           => sampleId + "."
-      case _                             => ""
+      case (Some(sampleId), Some(libId)) => sampleId + "-" + libId
+      case (Some(sampleId), _)           => sampleId
+      case _                             => config("name_prefix")
     }
   }
 
   def init: Unit = {
   }
 
-  def finalFile = new File(outputDir, namePrefix + "final.vcf.gz")
+  def finalFile = new File(outputDir, namePrefix + ".final.vcf.gz")
 
   def biopetScript: Unit = {
     val callers = usedCallers.sortBy(_.prio)
@@ -44,6 +44,12 @@ trait ShivaVariantcallingTrait extends SummaryQScript with SampleLibraryTag {
     for (caller <- callers) {
       caller.addJobs()
       cv.addInput(caller.outputFile, caller.name)
+
+      val vcfStats = new VcfStats(qscript)
+      vcfStats.input = caller.outputFile
+      vcfStats.setOutputDir(new File(caller.outputDir, "vcfstats"))
+      add(vcfStats)
+      addSummarizable(vcfStats, namePrefix + "-vcfstats-" + caller.name)
     }
     add(cv)
 
@@ -51,6 +57,7 @@ trait ShivaVariantcallingTrait extends SummaryQScript with SampleLibraryTag {
     vcfStats.input = finalFile
     vcfStats.setOutputDir(new File(outputDir, "vcfstats"))
     add(vcfStats)
+    addSummarizable(vcfStats, namePrefix + "-vcfstats-final")
 
     addSummaryJobs
   }
@@ -75,7 +82,7 @@ trait ShivaVariantcallingTrait extends SummaryQScript with SampleLibraryTag {
     protected val defaultPrio = 999
     protected val defaultUse = true
 
-    def outputFile = new File(outputDir, namePrefix + "raw.vcf.gz")
+    def outputFile = new File(outputDir, namePrefix + ".raw.vcf.gz")
 
     def addJobs() {
       val rawFiles = inputBams.map(bamFile => {