Adding tests for BamUtils

public/biopet-utils/src/main/scala/nl/lumc/sasc/biopet/utils/BamUtils.scala

@@ -48,7 +48,7 @@ object BamUtils {
     val insertsizes: List[Int] = (for {
       read <- samIterator.toStream.takeWhile(rec => {
         val paired = rec.getReadPairedFlag && rec.getProperPairFlag
-        val bothMapped = (rec.getReadUnmappedFlag == false) && (rec.getMateUnmappedFlag == false)
+        val bothMapped = if (paired) ((rec.getReadUnmappedFlag == false) && (rec.getMateUnmappedFlag == false)) else false
         paired && bothMapped
       }).take(samplingSize)
     } yield {
@@ -74,9 +74,12 @@ object BamUtils {
       contig => BamUtils.contigInsertSize(bamFile, contig.getSequenceName, 1, contig.getSequenceLength, samplingSize)
     }).toList
     val counts = baminsertsizes.flatMap(x => x)
-    val sum = counts.reduceLeft(_ + _)
-    val n = counts.size
-    sum / n
+
+    if (counts.size != 0) {
+      counts.reduceLeft(_ + _) / counts.size
+    } else {
+      0
+    }
   }
 
   /**

public/biopet-utils/src/test/scala/nl/lumc/sasc/biopet/utils/BamUtilsTest.scala

+package nl.lumc.sasc.biopet.utils
+
+import java.io.File
+
+import htsjdk.samtools._
+import org.mockito.Mockito.{inOrder => inOrd}
+import org.scalatest.Matchers
+import org.scalatest.mock.MockitoSugar
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.{BeforeClass, Test}
+
+/**
+ * Created by wyleung on 22-2-16.
+ * Create samfile and records are borrowed from WipeReadsTest by @bow
+ */
+class BamUtilsTest extends TestNGSuite with MockitoSugar with Matchers {
+
+  import BamUtils._
+
+  private val samHeaderTemplate: SAMLineParser = {
+    val samh = new SAMFileHeader
+    samh.setSortOrder(SAMFileHeader.SortOrder.coordinate)
+    samh.addSequence(new SAMSequenceRecord("chrQ", 10000))
+    samh.addSequence(new SAMSequenceRecord("chrR", 10000))
+    val readGroup1 = new SAMReadGroupRecord("001")
+    readGroup1.setSample("sample01")
+    samh.addReadGroup(readGroup1)
+    val readGroup2 = new SAMReadGroupRecord("002")
+    readGroup2.setSample("sample01")
+    samh.addReadGroup(readGroup2)
+    new SAMLineParser(samh)
+  }
+
+  private def makeSams(raws: String*): Seq[SAMRecord] =
+    raws.map(s => samHeaderTemplate.parseLine(s))
+
+  private def makeSamReader(f: File): SamReader = SamReaderFactory
+    .make()
+    .validationStringency(ValidationStringency.LENIENT)
+    .open(f)
+
+  val sBamRecs1 = makeSams(
+    "r02\t0\tchrQ\t50\t60\t10M\t*\t0\t0\tTACGTACGTA\tEEFFGGHHII\tRG:Z:001",
+    "r01\t16\tchrQ\t190\t60\t10M\t*\t0\t0\tTACGTACGTA\tEEFFGGHHII\tRG:Z:001",
+    "r01\t16\tchrQ\t290\t60\t10M\t*\t0\t0\tGGGGGAAAAA\tGGGGGGGGGG\tRG:Z:001",
+    "r04\t0\tchrQ\t450\t60\t10M\t*\t0\t0\tCGTACGTACG\tEEFFGGHHII\tRG:Z:001",
+    "r03\t16\tchrQ\t690\t60\t10M\t*\t0\t0\tCCCCCTTTTT\tHHHHHHHHHH\tRG:Z:001",
+    "r05\t0\tchrQ\t890\t60\t5M200N5M\t*\t0\t0\tGATACGATAC\tFEFEFEFEFE\tRG:Z:001",
+    "r06\t4\t*\t0\t0\t*\t*\t0\t0\tATATATATAT\tHIHIHIHIHI\tRG:Z:001"
+  )
+
+  val pBamRecs1 = makeSams(
+    "r02\t99\tchrQ\t50\t60\t10M\t=\t90\t50\tTACGTACGTA\tEEFFGGHHII\tRG:Z:001",
+    "r02\t147\tchrQ\t90\t60\t10M\t=\t50\t-50\tATGCATGCAT\tEEFFGGHHII\tRG:Z:001",
+    "r01\t163\tchrQ\t150\t60\t10M\t=\t190\t50\tAAAAAGGGGG\tGGGGGGGGGG\tRG:Z:001",
+    "r01\t83\tchrQ\t190\t60\t10M\t=\t150\t-50\tGGGGGAAAAA\tGGGGGGGGGG\tRG:Z:001",
+    "r01\t163\tchrQ\t250\t60\t10M\t=\t290\t50\tAAAAAGGGGG\tGGGGGGGGGG\tRG:Z:001",
+    "r01\t83\tchrQ\t290\t60\t10M\t=\t250\t-50\tGGGGGAAAAA\tGGGGGGGGGG\tRG:Z:001",
+    "r04\t99\tchrQ\t450\t60\t10M\t=\t490\t50\tCGTACGTACG\tEEFFGGHHII\tRG:Z:001",
+    "r04\t147\tchrQ\t490\t60\t10M\t=\t450\t-50\tGCATGCATGC\tEEFFGGHHII\tRG:Z:001",
+    "r03\t163\tchrQ\t650\t60\t10M\t=\t690\t50\tTTTTTCCCCC\tHHHHHHHHHH\tRG:Z:001",
+    "r03\t83\tchrQ\t690\t60\t10M\t=\t650\t-50\tCCCCCTTTTT\tHHHHHHHHHH\tRG:Z:001",
+    "r05\t99\tchrQ\t890\t60\t5M200N5M\t=\t1140\t50\tTACGTACGTA\tEEFFGGHHII\tRG:Z:001",
+    "r05\t147\tchrQ\t1140\t60\t10M\t=\t890\t-50\tATGCATGCAT\tEEFFGGHHII\tRG:Z:001",
+    "r06\t4\t*\t0\t0\t*\t*\t0\t0\tATATATATAT\tHIHIHIHIHI\tRG:Z:001",
+    "r06\t4\t*\t0\t0\t*\t*\t0\t0\tGCGCGCGCGC\tHIHIHIHIHI\tRG:Z:001"
+  )
+
+  @BeforeClass
+  def start: Unit = {
+    createTestFileFrom(sBamRecs1, BamUtilsTest.singleEndBam01)
+    createTestFileFrom(pBamRecs1, BamUtilsTest.pairedEndBam01)
+  }
+
+  private def createTestFileFrom(records: Seq[SAMRecord], output: File): File = {
+    output.getParentFile.createNewFile()
+    val outBam = new SAMFileWriterFactory()
+      .setCreateIndex(true)
+      .setUseAsyncIo(true)
+      .makeBAMWriter(samHeaderTemplate.getFileHeader, true, output)
+    writeBam(records, outBam)
+    output
+  }
+
+  private def writeBam(records: Seq[SAMRecord], outBam: SAMFileWriter): Unit = {
+    try {
+      for (rec <- records) {
+        outBam.addAlignment(rec)
+      }
+    } finally {
+      outBam.close()
+    }
+  }
+
+  @Test def testInputSingleEndOK() = {
+    sBamRecs1.size shouldBe 7
+  }
+
+  @Test def testInputPairedEndOK() = {
+    pBamRecs1.size shouldBe 14
+  }
+
+  @Test def testPairedRecords = {
+    sampleBamInsertSize(BamUtilsTest.pairedEndBam01) shouldBe 50
+  }
+
+  @Test def testContigInsertsize = {
+    contigInsertSize(BamUtilsTest.pairedEndBam01, "chrQ", 1, 10000) shouldBe Some(50)
+    contigInsertSize(BamUtilsTest.pairedEndBam01, "chrR", 1, 10000) shouldBe None
+  }
+
+  @Test def testContigInsertsizeContigNotFound = {
+    intercept(contigInsertSize(BamUtilsTest.pairedEndBam01, "chrW", 1, 10000))
+
+  }
+
+  @Test def testSingleRecords = {
+    sampleBamInsertSize(BamUtilsTest.singleEndBam01) shouldBe 0
+  }
+
+  @Test def testMultiBamInsertsizes = {
+    val result = sampleBamsInsertSize(List(BamUtilsTest.singleEndBam01, BamUtilsTest.pairedEndBam01))
+    result shouldEqual Map(
+      BamUtilsTest.singleEndBam01 -> 0,
+      BamUtilsTest.pairedEndBam01 -> 50
+    )
+  }
+
+  @Test def testSampleBamNames = {
+    sampleBamMap(List(BamUtilsTest.singleEndBam01)) shouldEqual Map(
+      "sample01" -> BamUtilsTest.singleEndBam01
+    )
+  }
+
+}
+
+object BamUtilsTest {
+  var singleEndBam01: File = File.createTempFile("bamutils", "single01.bam")
+  singleEndBam01.deleteOnExit()
+  var pairedEndBam01: File = File.createTempFile("bamutils", "paired01.bam")
+  singleEndBam01.deleteOnExit()
+}
\ No newline at end of file