Demultiplexed FASTQ support

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

@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+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] {
+
+  private def colBarcodeOpt = parser.indexBarcode.map(bc => FoundBarcode(bc.toCharArray, 0))
+
+  private[this] class BarcodeIterator(iterator: CloseableIterator[Read]) extends CloseableIterator[Barcodes] {
+    override def hasNext: Boolean = iterator.hasNext
+
+    override def next(): Barcodes = {
+      val nextRead = iterator.next()
+      val rowBarcodeOpt = rowPolicy.find(nextRead)
+      val umiBarcodeOpt = umiPolicyOpt.flatMap(_.find(nextRead))
+      Barcodes(rowBarcodeOpt, None, colBarcodeOpt, umiBarcodeOpt)
+    }
+
+    override def close(): Unit = iterator.close()
+  }
+
+  override def iterator: CloseableIterator[Barcodes] = new BarcodeIterator(parser.iterator)
+
+}

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

@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+package org.broadinstitute.gpp.poolq3.barcode
+
+import org.broadinstitute.gpp.poolq3.parser.{CloseableIterable, CloseableIterator, DmuxedIterable}
+import org.broadinstitute.gpp.poolq3.types.{Read, ReadIdCheckPolicy}
+
+class DmuxedPairedEndBarcodeSource(
+  rowParser: DmuxedIterable,
+  revRowParser: DmuxedIterable,
+  rowPolicy: BarcodePolicy,
+  revRowPolicy: BarcodePolicy,
+  umiPolicyOpt: Option[BarcodePolicy],
+  readIdCheckPolicy: ReadIdCheckPolicy
+) extends CloseableIterable[Barcodes] {
+
+  private def colBarcodeOpt = rowParser.indexBarcode.map(bc => FoundBarcode(bc.toCharArray, 0))
+
+  private[this] class BarcodeIterator(rowIterator: CloseableIterator[Read], revRowIterator: CloseableIterator[Read])
+      extends CloseableIterator[Barcodes] {
+
+    final override def hasNext: Boolean = rowIterator.hasNext && revRowIterator.hasNext
+
+    final override def next(): Barcodes = {
+      val nextRow = rowIterator.next()
+      val nextRevRow = revRowIterator.next()
+      readIdCheckPolicy.check(nextRow, nextRevRow)
+      val rowBarcodeOpt = rowPolicy.find(nextRow)
+      val revRowBarcodeOpt = revRowPolicy.find(nextRevRow)
+      val umiBarcodeOpt = umiPolicyOpt.flatMap(_.find(nextRow))
+      Barcodes(rowBarcodeOpt, revRowBarcodeOpt, colBarcodeOpt, umiBarcodeOpt)
+    }
+
+    final override def close(): Unit =
+      try rowIterator.close()
+      finally revRowIterator.close()
+
+  }
+
+  override def iterator: CloseableIterator[Barcodes] =
+    new BarcodeIterator(rowParser.iterator, revRowParser.iterator)
+
+}

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

@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+package org.broadinstitute.gpp.poolq3.barcode
+
+import cats.syntax.all._
+import munit.FunSuite
+import org.broadinstitute.gpp.poolq3.parser.DmuxedIterable
+
+class DmuxedBarcodeSourceTest extends FunSuite {
+
+  private[this] val rowPolicy = BarcodePolicy("FIXED@0", 10, skipShortReads = true)
+
+  def fb(s: String) = Barcodes(FoundBarcode(s.toCharArray, 0).some, None, None, None)
+
+  def fb(i: String, s: String) =
+    Barcodes(FoundBarcode(s.toCharArray, 0).some, None, FoundBarcode(i.toCharArray, 0).some, None)
+
+  test("it works") {
+    val iterable = DmuxedIterable(
+      List(
+        None -> List("AAAAAAAAAA", "AAAAAAAAAC", "AAAAAAAAAG"),
+        Some("CTCGAG") -> List("AAAAAAAAAA", "AACCCCGGTT", "AATTGGTTAA")
+      )
+    )
+
+    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None)
+    assertEquals(
+      src.toList,
+      List(
+        fb("AAAAAAAAAA"),
+        fb("AAAAAAAAAC"),
+        fb("AAAAAAAAAG"),
+        fb("CTCGAG", "AAAAAAAAAA"),
+        fb("CTCGAG", "AACCCCGGTT"),
+        fb("CTCGAG", "AATTGGTTAA")
+      )
+    )
+  }
+
+  test("nothing works") {
+    val iterable = DmuxedIterable(Nil)
+    val src = new DmuxedBarcodeSource(iterable, rowPolicy, None)
+    assertEquals(src.toList, Nil)
+  }
+
+}

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

@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2022 The Broad Institute, Inc. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+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
+
+class DmuxedPairedEndBarcodeSourceTest extends FunSuite {
+
+  private[this] val rowPolicy = BarcodePolicy("FIXED@0", 4, skipShortReads = true)
+  private[this] val revRowPolicy = BarcodePolicy("FIXED@0", 3, skipShortReads = true)
+
+  def fb(r1: String, r2: String) =
+    Barcodes(FoundBarcode(r1.toCharArray, 0).some, FoundBarcode(r2.toCharArray, 0).some, None, None)
+
+  def fb(i: String, r1: String, r2: String) =
+    Barcodes(
+      FoundBarcode(r1.toCharArray, 0).some,
+      FoundBarcode(r2.toCharArray, 0).some,
+      FoundBarcode(i.toCharArray, 0).some,
+      None
+    )
+
+  test("it works") {
+    val iter1 =
+      DmuxedIterable(List(None -> List("AAAA", "CCCC", "GGGG"), Some("CTCGAG") -> List("TTAA", "CCGG", "AATT")))
+
+    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)
+    assertEquals(
+      src.toList,
+      List(
+        fb("AAAA", "AGA"),
+        fb("CCCC", "CTC"),
+        fb("GGGG", "GAG"),
+        fb("CTCGAG", "TTAA", "TGT"),
+        fb("CTCGAG", "CCGG", "CAC"),
+        fb("CTCGAG", "AATT", "TCT")
+      )
+    )
+  }
+
+  test("nothing works") {
+    val i1 = DmuxedIterable(Nil)
+    val i2 = DmuxedIterable(Nil)
+    val src = new DmuxedPairedEndBarcodeSource(i1, i2, rowPolicy, revRowPolicy, None, ReadIdCheckPolicy.Illumina)
+    assertEquals(src.toList, Nil)
+  }
+
+}