The unmatched barcode incorrectly had a hyphen when multiple sample b…

…arcode reads were being used. (#317)
fulcrumgenomics · Nov 15, 2017 · e2d1d8d · e2d1d8d
1 parent 0841a6b
commit e2d1d8d
Show file tree

Hide file tree

Showing 2 changed files with 59 additions and 3 deletions.
diff --git a/src/main/scala/com/fulcrumgenomics/fastq/DemuxFastqs.scala b/src/main/scala/com/fulcrumgenomics/fastq/DemuxFastqs.scala
@@ -112,7 +112,7 @@ object DemuxFastqs {
     require(barcodeSegments.nonEmpty, "No sample barcodes found in read structures: " + readStructures.mkString(", "))
     require(barcodeSegments.forall(_.hasFixedLength), "Barcode segments must have fixed lengths in: " + readStructures.mkString(", "))
 
-    val noMatchBarcode: String = barcodeSegments.map("N" * _.fixedLength).mkString("-")
+    val noMatchBarcode: String = barcodeSegments.map("N" * _.fixedLength).mkString
     Sample(sampleOrdinal=sampleOrdinal, sampleId=UnmatchedSampleId, sampleName=UnmatchedSampleId, libraryId=UnmatchedSampleId, i7IndexBases=Some(noMatchBarcode))
   }
 
@@ -455,7 +455,7 @@ private class SamRecordWriter(output: PathToBam,
     }
     rec.sampleBarcode.foreach(bc => record("BC") = bc)
     record(ReservedTagConstants.READ_GROUP_ID) =  rgId
-    rec.molecularBarcode.foreach(mb => record(umiTag) =  mb)
+    rec.molecularBarcode.foreach(mb => record(umiTag) = mb)
     writer += record
   }
 

diff --git a/src/test/scala/com/fulcrumgenomics/fastq/DemuxFastqsTest.scala b/src/test/scala/com/fulcrumgenomics/fastq/DemuxFastqsTest.scala
@@ -27,7 +27,7 @@ package com.fulcrumgenomics.fastq
 
 import java.nio.file.Files
 
-import com.fulcrumgenomics.FgBioDef.{DirPath, FilePath}
+import com.fulcrumgenomics.FgBioDef.{DirPath, FilePath, PathToFastq}
 import com.fulcrumgenomics.fastq.FastqDemultiplexer.{DemuxRecord, DemuxResult}
 import com.fulcrumgenomics.illumina.{Sample, SampleSheet}
 import com.fulcrumgenomics.testing.{ErrorLogLevel, UnitSpec}
@@ -614,6 +614,62 @@ class DemuxFastqsTest extends UnitSpec with OptionValues with ErrorLogLevel {
     }
   }
 
+  it should "demux dual-indexed paired end reads" in {
+    def write(rec: FastqRecord): PathToFastq = {
+      val path = makeTempFile("fastq", ".fastq")
+      val writer = FastqWriter(path)
+      writer.write(rec)
+      writer.close()
+      path
+    }
+
+    val fq1 = write(fq(name="frag", bases="AAAAAAAA"))
+    val fq2 = write(fq(name="frag", bases="A"*100))
+    val fq3 = write(fq(name="frag", bases="T"*100))
+    val fq4 = write(fq(name="frag", bases="GATTACAGA"))
+    val structures = Seq(ReadStructure("8B"), ReadStructure("100T"), ReadStructure("100T"), ReadStructure("9B"))
+
+    val output: DirPath = outputDir()
+    val metrics = makeTempFile("metrics", ".txt")
+
+    new DemuxFastqs(inputs=Seq(fq1, fq2, fq3, fq4), output=output, metadata=sampleSheetPath,
+      readStructures=structures, metrics=Some(metrics), maxMismatches=2, minMismatchDelta=3).execute()
+
+    val sampleBarcodMetrics = Metric.read[SampleBarcodeMetric](metrics)
+    val sampleInfos = toSampleInfos(structures)
+    val samples: Seq[Sample] = sampleInfos.map(_.sample)
+
+    sampleBarcodMetrics.zip(samples).foreach { case (metric, sample) =>
+      metric.barcode_name shouldBe sample.sampleName
+      metric.barcode      shouldBe sample.sampleBarcodeString
+      metric.library_name shouldBe sample.libraryId
+      if (sample.sampleOrdinal == 1) {
+        metric.reads                   shouldBe 1
+        metric.pf_reads                shouldBe 1
+        metric.perfect_matches         shouldBe 1
+        metric.one_mismatch_matches    shouldBe 0
+        metric.pf_perfect_matches      shouldBe 1
+        metric.pf_one_mismatch_matches shouldBe 0
+      }
+      else if (sample.sampleId == UnmatchedSampleId) {
+        metric.reads                   shouldBe 0
+        metric.pf_reads                shouldBe 0
+        metric.perfect_matches         shouldBe 0
+        metric.one_mismatch_matches    shouldBe 0
+        metric.pf_perfect_matches      shouldBe 0
+        metric.pf_one_mismatch_matches shouldBe 0
+      }
+      else {
+        metric.reads                   shouldBe 0
+        metric.pf_reads                shouldBe 0
+        metric.perfect_matches         shouldBe 0
+        metric.one_mismatch_matches    shouldBe 0
+        metric.pf_perfect_matches      shouldBe 0
+        metric.pf_one_mismatch_matches shouldBe 0
+      }
+    }
+  }
+
   "FastqRecordWriter.readName" should "set the read name based if it should follow Illumina standards" in {
     val output = outputDir()
     val baseRec = DemuxRecord(name="name", bases="", quals="", molecularBarcode=Some("MB"), sampleBarcode=Some("SB"), readNumber=1, pairedEnd=false, comment=None)