From 955b7594b48d4afc9ef54201c930168112998280 Mon Sep 17 00:00:00 2001
From: znorgaard <zach@fulcrumgenomics.com>
Date: Tue, 23 Jul 2024 11:56:00 -0700
Subject: [PATCH] refactor: move isConsensus check

---
 .../umi/CollectDuplexSeqMetrics.scala             |  5 +++++
 .../umi/CollectDuplexSeqMetricsTest.scala         | 15 +++++++++++++--
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/src/main/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetrics.scala b/src/main/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetrics.scala
index 5d18b6d98..4d32a53bf 100644
--- a/src/main/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetrics.scala
+++ b/src/main/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetrics.scala
@@ -307,6 +307,11 @@ class CollectDuplexSeqMetrics
     // Build the iterator we'll use based on whether or not we're restricting to a set of intervals
     val in = SamSource(input)
     val _filteredIterator = in.iterator.filter(r => r.paired && r.mapped && r.mateMapped && r.firstOfPair && !r.secondary && !r.supplementary)
+      .tapEach {
+        r =>
+          if (Umis.isConsensusRead(r)) throw new IllegalArgumentException("Found consensus record. Expected UMI-grouped BAM")
+          else r
+      }
     val iterator = intervals match {
       case None       => _filteredIterator
       case Some(path) =>
diff --git a/src/test/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetricsTest.scala b/src/test/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetricsTest.scala
index f1939c68d..8c646a91a 100644
--- a/src/test/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetricsTest.scala
+++ b/src/test/scala/com/fulcrumgenomics/umi/CollectDuplexSeqMetricsTest.scala
@@ -26,13 +26,13 @@ package com.fulcrumgenomics.umi
 
 import java.nio.file.{Path, Paths}
 import java.util.Random
-
 import com.fulcrumgenomics.FgBioDef._
 import com.fulcrumgenomics.bam.api.SamOrder
 import com.fulcrumgenomics.commons.util.SimpleCounter
 import com.fulcrumgenomics.testing.SamBuilder.{Minus, Plus}
 import com.fulcrumgenomics.testing.{SamBuilder, UnitSpec}
-import com.fulcrumgenomics.umi.ConsensusTags.PerRead.AllPerReadTags
+import com.fulcrumgenomics.umi.ConsensusTags.PerBase.AbRawReadCount
+import com.fulcrumgenomics.umi.ConsensusTags.PerRead.{AllPerReadTags, BaRawReadCount}
 import com.fulcrumgenomics.util.{Io, Metric, Rscript}
 import htsjdk.samtools.util.{Interval, IntervalList}
 import org.apache.commons.math3.distribution.NormalDistribution
@@ -314,6 +314,17 @@ class CollectDuplexSeqMetricsTest extends UnitSpec {
     duplexFamilies.find(f => f.ab_size == 6 && f.ba_size == 0).get.count shouldBe 1
   }
 
+  it should "raise an exception if duplex consensus records are provided" in {
+    val builder = new SamBuilder(readLength=100, sort=Some(SamOrder.TemplateCoordinate))
+    builder.addPair(
+      contig=1,
+      start1=1000,
+      start2=1100,
+      attrs=Map(RX -> "AAA-GGG", MI -> "1/A", AbRawReadCount -> 10, BaRawReadCount -> 10)
+    )
+    an[IllegalArgumentException] shouldBe thrownBy { exec(builder) }
+  }
+
   "CollectDuplexSeqMetrics.updateUmiMetrics" should "not count duplex umis" in collector(duplex=false).foreach { c =>
     val builder = new SamBuilder(readLength=10)
     builder.addPair(start1=100, start2=200, attrs=Map(RX -> "AAA-CCC", MI -> "1/A"))