Adding validateVcf to shiva on dbsnp

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

+/**
+ * Biopet is built on top of GATK Queue for building bioinformatic
+ * pipelines. It is mainly intended to support LUMC SHARK cluster which is running
+ * SGE. But other types of HPC that are supported by GATK Queue (such as PBS)
+ * should also be able to execute Biopet tools and pipelines.
+ *
+ * Copyright 2014 Sequencing Analysis Support Core - Leiden University Medical Center
+ *
+ * Contact us at: sasc@lumc.nl
+ *
+ * A dual licensing mode is applied. The source code within this project is freely available for non-commercial use under an AGPL
+ * license; For commercial users or users who do not want to follow the AGPL
+ * license, please contact us to obtain a separate license.
+ */
+package nl.lumc.sasc.biopet.extensions.tools
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.{Reference, ToolCommandFunction}
+import nl.lumc.sasc.biopet.utils.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.Input
+
+class ValidateVcf(val root: Configurable) extends ToolCommandFunction with Reference {
+  def toolObject = nl.lumc.sasc.biopet.tools.ValidateVcf
+
+  @Input(required = true)
+  var inputVcf: File = _
+
+  @Input(required = true)
+  var reference: File = _
+
+  var disableFail: Boolean = false
+
+  override def defaultCoreMemory = 4.0
+
+  override def beforeGraph(): Unit = {
+    super.beforeGraph()
+    if (reference == null) reference = referenceFasta()
+  }
+
+  override def cmdLine = super.cmdLine +
+    required("-i", inputVcf) +
+    required("-R", reference) +
+    conditional(disableFail, "--disableFail")
+}

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/CheckValidateVcf.scala

+/**
+ * Biopet is built on top of GATK Queue for building bioinformatic
+ * pipelines. It is mainly intended to support LUMC SHARK cluster which is running
+ * SGE. But other types of HPC that are supported by GATK Queue (such as PBS)
+ * should also be able to execute Biopet tools and pipelines.
+ *
+ * Copyright 2014 Sequencing Analysis Support Core - Leiden University Medical Center
+ *
+ * Contact us at: sasc@lumc.nl
+ *
+ * A dual licensing mode is applied. The source code within this project is freely available for non-commercial use under an AGPL
+ * license; For commercial users or users who do not want to follow the AGPL
+ * license, please contact us to obtain a separate license.
+ */
+package nl.lumc.sasc.biopet.pipelines.shiva
+
+import java.io.File
+
+import org.broadinstitute.gatk.queue.function.InProcessFunction
+import org.broadinstitute.gatk.utils.commandline.Input
+
+import scala.io.Source
+
+/**
+ * This class checks md5sums and give an exit code 1 when md5sum is not the same
+ *
+ * Created by pjvanthof on 16/08/15.
+ */
+class CheckValidateVcf extends InProcessFunction {
+  @Input(required = true)
+  var inputLogFile: File = _
+
+  /** Exits whenever the input md5sum is not the same as the output md5sum */
+  def run: Unit = {
+
+    val reader = Source.fromFile(inputLogFile)
+    reader.getLines().foreach { line =>
+      if (line.startsWith("ERROR")) {
+        logger.error("Corrupt vcf file found, aborting pipeline")
+
+        // 130 Simulates a ctr-C
+        Runtime.getRuntime.halt(130)
+      }
+    }
+    reader.close()
+  }
+}
\ No newline at end of file

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/Shiva.scala

@@ -14,15 +14,19 @@
  */
 package nl.lumc.sasc.biopet.pipelines.shiva
 
-import nl.lumc.sasc.biopet.core.{ PipelineCommand, Reference }
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.{PipelineCommand, Reference}
 import nl.lumc.sasc.biopet.core.report.ReportBuilderExtension
 import nl.lumc.sasc.biopet.extensions.gatk._
+import nl.lumc.sasc.biopet.extensions.tools.ValidateVcf
 import nl.lumc.sasc.biopet.pipelines.bammetrics.TargetRegions
 import nl.lumc.sasc.biopet.pipelines.kopisu.Kopisu
 import nl.lumc.sasc.biopet.pipelines.mapping.MultisampleMappingTrait
 import nl.lumc.sasc.biopet.pipelines.toucan.Toucan
 import nl.lumc.sasc.biopet.utils.config.Configurable
 import org.broadinstitute.gatk.queue.QScript
+import org.broadinstitute.gatk.queue.function.QFunction
 
 /**
  * This is a trait for the Shiva pipeline
@@ -175,6 +179,8 @@ class Shiva(val root: Configurable) extends QScript with MultisampleMappingTrait
   override def addMultiSampleJobs() = {
     super.addMultiSampleJobs()
 
+    addAll(dbsnpVcfFile.map(Shiva.makeValidateVcfJobs(this, _, referenceFasta())).getOrElse(Nil))
+
     multisampleVariantCalling.foreach(vc => {
       vc.outputDir = new File(outputDir, "variantcalling")
       vc.inputBams = samples.flatMap { case (sampleId, sample) => sample.preProcessBam.map(sampleId -> _) }
@@ -251,4 +257,23 @@ class Shiva(val root: Configurable) extends QScript with MultisampleMappingTrait
 }
 
 /** This object give a default main method to the pipelines */
-object Shiva extends PipelineCommand
\ No newline at end of file
+object Shiva extends PipelineCommand {
+
+  // This is used to only execute 1 validation per vcf file
+  private var validateVcfSeen: Set[(File, File)] = Set()
+
+  def makeValidateVcfJobs(root: Configurable, vcfFile: File, referenceFile: File): List[QFunction] = {
+    if (validateVcfSeen.contains((vcfFile, referenceFile))) Nil
+    else {
+      validateVcfSeen ++= Set((vcfFile, referenceFile))
+      val validateVcf = new ValidateVcf(root)
+      validateVcf.inputVcf = vcfFile
+      validateVcf.reference = referenceFile
+
+      val checkValidateVcf = new CheckValidateVcf
+      checkValidateVcf.inputLogFile = validateVcf.jobOutputFile
+
+      List(validateVcf, checkValidateVcf)
+    }
+  }
+}
\ No newline at end of file

shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/ShivaVariantcalling.scala

@@ -85,6 +85,8 @@ class ShivaVariantcalling(val root: Configurable) extends QScript
     require(inputBams.nonEmpty, "No input bams found")
     require(callers.nonEmpty, "must select at least 1 variantcaller, choices are: " + callersList.map(_.name).mkString(", "))
 
+    addAll(dbsnpVcfFile.map(Shiva.makeValidateVcfJobs(this, _, referenceFasta())).getOrElse(Nil))
+
     val cv = new CombineVariants(qscript)
     cv.out = finalFile
     cv.setKey = Some("VariantCaller")