Added base version of validate vcf

biopet-tools-package/src/main/scala/nl/lumc/sasc/biopet/BiopetToolsExecutable.scala

@@ -45,6 +45,7 @@ object BiopetToolsExecutable extends BiopetExecutable {
     nl.lumc.sasc.biopet.tools.SquishBed,
     nl.lumc.sasc.biopet.tools.SummaryToTsv,
     nl.lumc.sasc.biopet.tools.ValidateFastq,
+    nl.lumc.sasc.biopet.tools.ValidateVcf,
     nl.lumc.sasc.biopet.tools.VcfFilter,
     nl.lumc.sasc.biopet.tools.VcfStats,
     nl.lumc.sasc.biopet.tools.VcfToTsv,

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

+package nl.lumc.sasc.biopet.tools
+
+import java.io.File
+
+import htsjdk.variant.vcf.VCFFileReader
+import nl.lumc.sasc.biopet.utils.ToolCommand
+import nl.lumc.sasc.biopet.utils.intervals.{ BedRecord, BedRecordList }
+
+import scala.collection.JavaConversions._
+
+/**
+  * Created by pjvanthof on 10/12/2016.
+  */
+object ValidateVcf extends ToolCommand {
+  case class Args(inputVcf: File = null,
+                  reference: File = null,
+                  failOnError: Boolean = true) extends AbstractArgs
+
+  class OptParser extends AbstractOptParser {
+    opt[File]('i', "inputVcf") required () maxOccurs 1 valueName "<file>" action { (x, c) =>
+      c.copy(inputVcf = x)
+    } text "Vcf file to check"
+    opt[File]('R', "reference") required () maxOccurs 1 valueName "<file>" action { (x, c) =>
+      c.copy(reference = x)
+    } text "Reference fasta to check vcf file against"
+    opt[Unit]("disableFail") maxOccurs 1 valueName "<file>" action { (x, c) =>
+      c.copy(failOnError = false)
+    } text "Disable failing, tool still stops at the first error but exitcode 0 is given"
+  }
+
+  def main(args: Array[String]): Unit = {
+    logger.info("Start")
+
+    val argsParser = new OptParser
+    val cmdArgs: Args = argsParser.parse(args, Args()) getOrElse (throw new IllegalArgumentException)
+
+    val regions = BedRecordList.fromReference(cmdArgs.reference)
+
+    val vcfReader = new VCFFileReader(cmdArgs.inputVcf, true)
+
+    try {
+      for (record <- vcfReader.iterator()) {
+        val contig = record.getContig
+        require(regions.chrRecords.contains(contig),
+          s"contig in vcf file is not on reference: ${contig}")
+        val start = record.getStart
+        val end = record.getEnd
+        require(regions.chrRecords(contig).exists(_.overlapWith(BedRecord("contig", start, start))),
+          s"Position does not exist on reference: ${contig}:$start")
+        if (end != start) require(regions.chrRecords(contig).exists(_.overlapWith(BedRecord("contig", end, end))),
+          s"Position does not exist on reference: ${contig}:$end")
+        require(start <= end, "End is higher then Start, this should not be possible")
+      }
+    } catch {
+      case e: IllegalArgumentException =>
+        if (cmdArgs.failOnError) throw e
+        else logger.warn(e.getMessage)
+    }
+
+    vcfReader.close()
+
+    logger.info("No error found")
+  }
+}