Parse barcodes from read IDs in demultiplexed mode

src/main/scala/org/broadinstitute/gpp/poolq3/PoolQ.scala

@@ -116,8 +116,10 @@ object PoolQ {
       ExactReference(referenceData.mappings, identity, includeAmbiguous = false)
     }
 
+    val colBarcodePolicyOrLength: Either[Int, BarcodePolicy] = colBarcodePolicyOpt.toRight(colReference.barcodeLength)
+
     val barcodes: CloseableIterable[Barcodes] =
-      barcodeSource(config.input, rowBarcodePolicy, revRowBarcodePolicyOpt, colBarcodePolicyOpt, umiInfo.map(_._2))
+      barcodeSource(config.input, rowBarcodePolicy, revRowBarcodePolicyOpt, colBarcodePolicyOrLength, umiInfo.map(_._2))
 
     lazy val unexpectedSequenceCacheDir: Option[Path] =
       if (config.skipUnexpectedSequenceReport) None

src/main/scala/org/broadinstitute/gpp/poolq3/barcode/Dmuxed.scala

@@ -0,0 +1,18 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+package org.broadinstitute.gpp.poolq3.barcode
+
+object Dmuxed {
+
+  private[barcode] def barcodeFromId(length: Int): String => Option[FoundBarcode] = {
+    val regex = s"@.*[^ACGTN]([ACGTN]{$length})$$".r
+    _ match {
+      case regex(barcode) => Some(FoundBarcode(barcode.toCharArray, 0))
+      case _              => None
+    }
+  }
+
+}

src/main/scala/org/broadinstitute/gpp/poolq3/barcode/DmuxedBarcodeSource.scala

@@ -8,8 +8,15 @@ package org.broadinstitute.gpp.poolq3.barcode
 import org.broadinstitute.gpp.poolq3.parser.{CloseableIterable, CloseableIterator, DmuxedIterable}
 import org.broadinstitute.gpp.poolq3.types.Read
 
-final class DmuxedBarcodeSource(parser: DmuxedIterable, rowPolicy: BarcodePolicy, umiPolicyOpt: Option[BarcodePolicy])
-    extends CloseableIterable[Barcodes] {
+final class DmuxedBarcodeSource(
+  parser: DmuxedIterable,
+  rowPolicy: BarcodePolicy,
+  umiPolicyOpt: Option[BarcodePolicy],
+  colBarcodeLength: Int
+) extends CloseableIterable[Barcodes] {
+
+  // used to attempt to parse barcodes out of ids if the file has no associated barcode
+  private val colBarcodeParser = Dmuxed.barcodeFromId(colBarcodeLength)
 
   private def colBarcodeOpt = parser.indexBarcode
 
@@ -20,7 +27,7 @@ final class DmuxedBarcodeSource(parser: DmuxedIterable, rowPolicy: BarcodePolicy
       val nextRead = iterator.next()
       val rowBarcodeOpt = rowPolicy.find(nextRead)
       val umiBarcodeOpt = umiPolicyOpt.flatMap(_.find(nextRead))
-      Barcodes(rowBarcodeOpt, None, colBarcodeOpt, umiBarcodeOpt)
+      Barcodes(rowBarcodeOpt, None, colBarcodeOpt.orElse(colBarcodeParser(nextRead.id)), umiBarcodeOpt)
     }
 
     override def close(): Unit = iterator.close()

src/main/scala/org/broadinstitute/gpp/poolq3/barcode/DmuxedPairedEndBarcodeSource.scala

@@ -14,7 +14,8 @@ class DmuxedPairedEndBarcodeSource(
   rowPolicy: BarcodePolicy,
   revRowPolicy: BarcodePolicy,
   umiPolicyOpt: Option[BarcodePolicy],
-  readIdCheckPolicy: ReadIdCheckPolicy
+  readIdCheckPolicy: ReadIdCheckPolicy,
+  colBarcodeLength: Int
 ) extends CloseableIterable[Barcodes] {
 
   // the index barcode _is_ the column barcode; we get it from the row parser
@@ -24,6 +25,9 @@ class DmuxedPairedEndBarcodeSource(
   private[this] class BarcodeIterator(rowIterator: CloseableIterator[Read], revRowIterator: CloseableIterator[Read])
       extends CloseableIterator[Barcodes] {
 
+    // used to attempt to parse barcodes out of ids if the file has no associated barcode
+    private val colBarcodeParser = Dmuxed.barcodeFromId(colBarcodeLength)
+
     final override def hasNext: Boolean = rowIterator.hasNext && revRowIterator.hasNext
 
     final override def next(): Barcodes = {
@@ -33,7 +37,7 @@ class DmuxedPairedEndBarcodeSource(
       val rowBarcodeOpt = rowPolicy.find(nextRow)
       val revRowBarcodeOpt = revRowPolicy.find(nextRevRow)
       val umiBarcodeOpt = umiPolicyOpt.flatMap(_.find(nextRow))
-      Barcodes(rowBarcodeOpt, revRowBarcodeOpt, colBarcodeOpt, umiBarcodeOpt)
+      Barcodes(rowBarcodeOpt, revRowBarcodeOpt, colBarcodeOpt.orElse(colBarcodeParser(nextRow.id)), umiBarcodeOpt)
     }
 
     final override def close(): Unit =

src/main/scala/org/broadinstitute/gpp/poolq3/barcode/package.scala

@@ -26,11 +26,11 @@ package object barcode {
     config: PoolQInput,
     rowBarcodePolicy: BarcodePolicy,
     revRowBarcodePolicyOpt: Option[BarcodePolicy],
-    colBarcodePolicyOpt: Option[BarcodePolicy],
+    colBarcodePolicyOpt: Either[Int, BarcodePolicy],
     umiBarcodePolicyOpt: Option[BarcodePolicy]
   ): CloseableIterable[Barcodes] =
     (config.readsSource, revRowBarcodePolicyOpt, colBarcodePolicyOpt) match {
-      case (ReadsSource.Split(index, forward), None, Some(colBarcodePolicy)) =>
+      case (ReadsSource.Split(index, forward), None, Right(colBarcodePolicy)) =>
         new TwoFileBarcodeSource(
           parserFor(forward.toList),
           parserFor(index.toList),
@@ -39,7 +39,7 @@ package object barcode {
           umiBarcodePolicyOpt,
           config.readIdCheckPolicy
         )
-      case (ReadsSource.PairedEnd(index, forward, reverse), Some(revRowBarcodePolicy), Some(colBarcodePolicy)) =>
+      case (ReadsSource.PairedEnd(index, forward, reverse), Some(revRowBarcodePolicy), Right(colBarcodePolicy)) =>
         new ThreeFileBarcodeSource(
           parserFor(forward.toList),
           parserFor(reverse.toList),
@@ -50,22 +50,24 @@ package object barcode {
           umiBarcodePolicyOpt,
           config.readIdCheckPolicy
         )
-      case (ReadsSource.SelfContained(paths), None, Some(colBarcodePolicy)) =>
+      case (ReadsSource.SelfContained(paths), None, Right(colBarcodePolicy)) =>
         new SingleFileBarcodeSource(parserFor(paths.toList), rowBarcodePolicy, colBarcodePolicy, umiBarcodePolicyOpt)
-      case (ReadsSource.Dmuxed(read1), _, _) =>
+      case (ReadsSource.Dmuxed(read1), _, Left(colBarcodeLength)) =>
         new DmuxedBarcodeSource(
           DmuxedIterable(read1.toList, parserFor(_).iterator),
           rowBarcodePolicy,
-          umiBarcodePolicyOpt
+          umiBarcodePolicyOpt,
+          colBarcodeLength
         )
-      case (ReadsSource.DmuxedPairedEnd(read1, read2), Some(revRowBarcodePolicy), _) =>
+      case (ReadsSource.DmuxedPairedEnd(read1, read2), Some(revRowBarcodePolicy), Left(colBarcodeLength)) =>
         new DmuxedPairedEndBarcodeSource(
           DmuxedIterable(read1.toList, parserFor(_).iterator),
           DmuxedIterable(read2.toList, parserFor(_).iterator),
           rowBarcodePolicy,
           revRowBarcodePolicy,
           umiBarcodePolicyOpt,
-          config.readIdCheckPolicy
+          config.readIdCheckPolicy,
+          colBarcodeLength
         )
       case _ =>
         throw new IllegalArgumentException("Incompatible reads and barcode policy settings")

src/main/scala/org/broadinstitute/gpp/poolq3/parser/CloseableIterable.scala

@@ -50,9 +50,12 @@ object DmuxedIterable {
     val data2: List[(Option[String], List[Read])] = data.map { case (bco, seqs) =>
       (bco, seqs.zipWithIndex.map { case (seq, i) => Read(i.toString, seq) })
     }
-    new DmuxedIterableImpl(data2, CloseableIterator.ofList)
+    DmuxedIterable.forReads(data2)
   }
 
+  def forReads(data: List[(Option[String], List[Read])]): DmuxedIterable =
+    new DmuxedIterableImpl(data, CloseableIterator.ofList)
+
   private class DmuxedIterableImpl[A](src: Iterable[(Option[String], A)], makeIterator: A => CloseableIterator[Read])
       extends DmuxedIterable {
 

src/test/scala/org/broadinstitute/gpp/poolq3/barcode/DmuxedBarcodeSourceTest.scala

@@ -8,6 +8,7 @@ package org.broadinstitute.gpp.poolq3.barcode
 import cats.syntax.all._
 import munit.FunSuite
 import org.broadinstitute.gpp.poolq3.parser.DmuxedIterable
+import org.broadinstitute.gpp.poolq3.types.Read
 
 class DmuxedBarcodeSourceTest extends FunSuite {
 
@@ -26,7 +27,7 @@ class DmuxedBarcodeSourceTest extends FunSuite {
       )
     )
 
-    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None)
+    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None, 8)
     assertEquals(
       src.toList,
       List(
@@ -40,9 +41,26 @@ class DmuxedBarcodeSourceTest extends FunSuite {
     )
   }
 
+  test("barcodes from read IDs") {
+    val undeterminedReads = List(Read("@eeeeee ACGTAA", "AAAAAAAAAA"), Read("@eeeeee ACTCAG", "CCCCCCCCCC"))
+    val aacctgReads = List(Read("@a read", "GGGGGGGGGG"), Read("@another read", "TTTTTTTTTT"))
+    val iterable = DmuxedIterable.forReads(List(None -> undeterminedReads, Some("AACCTG") -> aacctgReads))
+
+    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None, 6)
+    assertEquals(
+      src.toList,
+      List(
+        fb("ACGTAA", "AAAAAAAAAA"),
+        fb("ACTCAG", "CCCCCCCCCC"),
+        fb("AACCTG", "GGGGGGGGGG"),
+        fb("AACCTG", "TTTTTTTTTT")
+      )
+    )
+  }
+
   test("nothing works") {
     val iterable = DmuxedIterable(Nil)
-    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None)
+    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None, 8)
     assertEquals(src.toList, Nil)
   }
 

src/test/scala/org/broadinstitute/gpp/poolq3/barcode/DmuxedPairedEndBarcodeSourceTest.scala

@@ -8,7 +8,7 @@ package org.broadinstitute.gpp.poolq3.barcode
 import cats.syntax.all._
 import munit.FunSuite
 import org.broadinstitute.gpp.poolq3.parser.DmuxedIterable
-import org.broadinstitute.gpp.poolq3.types.ReadIdCheckPolicy
+import org.broadinstitute.gpp.poolq3.types.{Read, ReadIdCheckPolicy}
 
 class DmuxedPairedEndBarcodeSourceTest extends FunSuite {
 
@@ -32,7 +32,7 @@ class DmuxedPairedEndBarcodeSourceTest extends FunSuite {
 
     val iter2 = DmuxedIterable(List(None -> List("AGA", "CTC", "GAG"), Some("CTCGAG") -> List("TGT", "CAC", "TCT")))
 
-    val src = new DmuxedPairedEndBarcodeSource(iter1, iter2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Lax)
+    val src = new DmuxedPairedEndBarcodeSource(iter1, iter2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Lax, 8)
     assertEquals(
       src.toList,
       List(
@@ -46,10 +46,31 @@ class DmuxedPairedEndBarcodeSourceTest extends FunSuite {
     )
   }
 
+  test("barcodes from read IDs") {
+    val undeterminedRead1s = List(Read("@eeeeee ACGTAA", "AAAA"), Read("@eeeeee ACTCAG", "CCCC"))
+    val undeterminedRead2s = List(Read("@eeeeee ACGTAA", "AAA"), Read("@eeeeee ACTCAG", "CCC"))
+    val aacctgRead1s = List(Read("@a read", "GGGG"), Read("@another read", "TTTT"))
+    val aacctgRead2s = List(Read("@a read", "GGG"), Read("@another read", "TTT"))
+
+    val iter1 = DmuxedIterable.forReads(List(None -> undeterminedRead1s, Some("AACCTG") -> aacctgRead1s))
+    val iter2 = DmuxedIterable.forReads(List(None -> undeterminedRead2s, Some("AACCTG") -> aacctgRead2s))
+
+    val src = new DmuxedPairedEndBarcodeSource(iter1, iter2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Lax, 6)
+    assertEquals(
+      src.toList,
+      List(
+        fb("ACGTAA", "AAAA", "AAA"),
+        fb("ACTCAG", "CCCC", "CCC"),
+        fb("AACCTG", "GGGG", "GGG"),
+        fb("AACCTG", "TTTT", "TTT")
+      )
+    )
+  }
+
   test("nothing works") {
     val i1 = DmuxedIterable(Nil)
     val i2 = DmuxedIterable(Nil)
-    val src = new DmuxedPairedEndBarcodeSource(i1, i2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Illumina)
+    val src = new DmuxedPairedEndBarcodeSource(i1, i2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Illumina, 8)
     assertEquals(src.toList, Nil)
   }
 

src/test/scala/org/broadinstitute/gpp/poolq3/barcode/DmuxedTest.scala

@@ -0,0 +1,34 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+package org.broadinstitute.gpp.poolq3.barcode
+
+import munit.FunSuite
+
+class DmuxedTest extends FunSuite {
+
+  test("extracting a barcode with Ns from an illumina read") {
+    assertEquals(
+      Dmuxed.barcodeFromId(8)("@A01379:680:HC37HDRX3:1:2101:1163:1000 1:N:0:GGNGNANT"),
+      Some(FoundBarcode("GGNGNANT".toCharArray, 0))
+    )
+  }
+
+  test("extracting a barcode with no Ns from an illumina read") {
+    assertEquals(
+      Dmuxed.barcodeFromId(8)("@A01379:680:HC37HDRX3:1:2101:3224:1000 1:N:0:AAATGCGA"),
+      Some(FoundBarcode("AAATGCGA".toCharArray, 0))
+    )
+  }
+
+  test("ignore a barcode-like sequence that's too long") {
+    assertEquals(Dmuxed.barcodeFromId(8)("@A01379:680:HC37HDRX3:1:2101:3224:1000 1:N:0:AAATGCGAGG"), None)
+  }
+
+  test("ignore a barcode-like sequence that's too short") {
+    assertEquals(Dmuxed.barcodeFromId(8)("@A01379:680:HC37HDRX3:1:2101:3224:1000 1:N:0:TGCGAGG"), None)
+  }
+
+}