Merge branch 'fix-must-have-variant' into 'develop'

Fix must have variant

Fixes #386 

See merge request !455

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

@@ -19,7 +19,7 @@ import java.io.File
 import htsjdk.variant.variantcontext.{ GenotypeType, VariantContext }
 import htsjdk.variant.variantcontext.writer.{ AsyncVariantContextWriter, VariantContextWriterBuilder }
 import htsjdk.variant.vcf.VCFFileReader
-import nl.lumc.sasc.biopet.utils.ToolCommand
+import nl.lumc.sasc.biopet.utils.{ ToolCommand, VcfUtils }
 
 import scala.collection.JavaConversions._
 import scala.io.Source
@@ -315,11 +315,11 @@ object VcfFilter extends ToolCommand {
    * Checks if given samples does have a variant hin this record
    *
    * @param record VCF record
-   * @param mustHaveVariant List of samples that should have this variant
+   * @param samples List of samples that should have this variant
    * @return true if filter passed
    */
-  def mustHaveVariant(record: VariantContext, mustHaveVariant: List[String]): Boolean = {
-    !mustHaveVariant.map(record.getGenotype).exists(a => a.isHomRef || a.isNoCall)
+  def mustHaveVariant(record: VariantContext, samples: List[String]): Boolean = {
+    !samples.map(record.getGenotype).exists(a => a.isHomRef || a.isNoCall || VcfUtils.isCompoundNoCall(a))
   }
 
   /** Checks if given samples have the same genotype */

biopet-tools/src/test/scala/nl/lumc/sasc/biopet/tools/VcfFilterTest.scala

@@ -39,7 +39,9 @@ class VcfFilterTest extends TestNGSuite with MockitoSugar with Matchers {
   }
 
   val veppedPath = resourcePath("/VEP_oneline.vcf")
+  val starPath = resourcePath("/star_genotype.vcf.gz")
   val vepped = new File(veppedPath)
+  val star = new File(starPath)
   val rand = new Random()
 
   @Test def testOutputTypeVcf() = {
@@ -183,6 +185,9 @@ class VcfFilterTest extends TestNGSuite with MockitoSugar with Matchers {
     mustHaveVariant(record, List("Sample_101")) shouldBe true
     mustHaveVariant(record, List("Sample_101", "Sample_102")) shouldBe true
     mustHaveVariant(record, List("Sample_101", "Sample_102", "Sample_103")) shouldBe false
+
+    val starReader = new VCFFileReader(star, false)
+    starReader.iterator().foreach(x => mustHaveVariant(x, List("Sample_101")) shouldBe false)
   }
 
   @Test def testSameGenotype() = {

biopet-utils/src/main/scala/nl/lumc/sasc/biopet/utils/VcfUtils.scala

@@ -140,4 +140,13 @@ object VcfUtils {
     reader.close()
     !hasNext
   }
+
+  /**
+   * Check whether genotype is of the form 0/.
+   * @param genotype genotype
+   * @return boolean
+   */
+  def isCompoundNoCall(genotype: Genotype): Boolean = {
+    genotype.isCalled && genotype.getAlleles.exists(_.isNoCall) && genotype.getAlleles.exists(_.isReference)
+  }
 }

biopet-utils/src/test/scala/VcfUtilsTest.scala

+import htsjdk.variant.variantcontext.{ Allele, Genotype, GenotypeBuilder }
+import org.scalatest.Matchers
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.collection.JavaConversions._
+
+import nl.lumc.sasc.biopet.utils.VcfUtils
+
+/**
+ * Created by Sander Bollen on 4-10-16.
+ */
+class VcfUtilsTest extends TestNGSuite with Matchers {
+
+  @Test
+  def testCompoundNoCall(): Unit = {
+    val noAllele = Allele.NO_CALL
+    val refAllele = Allele.create("A", true)
+    val compoundNoCall = GenotypeBuilder.create("sample_01", List(noAllele, refAllele))
+    VcfUtils.isCompoundNoCall(compoundNoCall) shouldBe true
+
+    val altAllele = Allele.create("G", false)
+    val normalGenotype = GenotypeBuilder.create("sample_01", List(refAllele, altAllele))
+    VcfUtils.isCompoundNoCall(normalGenotype) shouldBe false
+
+    val completeNoCall = GenotypeBuilder.create("sample_01", List(noAllele, noAllele))
+    VcfUtils.isCompoundNoCall(completeNoCall) shouldBe false
+  }
+
+}