From 90e19321bcca6f6266c541f4273e354d86907c38 Mon Sep 17 00:00:00 2001
From: Dror Kessler <dror27.kessler@gmail.com>
Date: Mon, 19 Jun 2023 10:14:23 +0300
Subject: [PATCH] --include-qc-failed-reads option

---
 .../featuremapping/FlowFeatureMapper.java     | 12 +++++++
 .../FlowFeatureMapperIntegrationTest.java     | 31 +++++++++++++++++++
 .../snv_feature_mapper_qc_failed_output.vcf   |  3 ++
 ...nv_feature_mapper_qc_failed_output.vcf.idx |  3 ++
 4 files changed, 49 insertions(+)
 create mode 100644 src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf
 create mode 100644 src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf.idx

diff --git a/src/main/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapper.java b/src/main/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapper.java
index b3ddc3e21b6..82ce79ad5d6 100644
--- a/src/main/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapper.java
+++ b/src/main/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapper.java
@@ -103,6 +103,9 @@ public final class FlowFeatureMapper extends ReadWalker {
     private static final String     VCF_SMQ_RIGHT_MEAN = "X_SMQ_RIGHT_MEAN";
 
     private static final Double     LOWEST_PROB = 0.0001;
+    private static final int        VENDOR_QUALITY_CHECK_FLAG = 0x200;
+
+    private static final String     INCLUDE_QC_FAILED_READS_FULL_NAME = "include-qc-failed-reads";
 
     @Argument(fullName = StandardArgumentDefinitions.OUTPUT_LONG_NAME,
             shortName = StandardArgumentDefinitions.OUTPUT_SHORT_NAME,
@@ -131,6 +134,10 @@ public final class FlowFeatureMapper extends ReadWalker {
     @Argument(fullName=HaplotypeCallerArgumentCollection.OUTPUT_BLOCK_LOWER_BOUNDS, doc = "Output the band lower bound for each GQ block regardless of the data it represents", optional = true)
     public boolean floorBlocks = false;
 
+    @Advanced
+    @Argument(fullName=INCLUDE_QC_FAILED_READS_FULL_NAME, doc = "include reads with QC failed flag", optional = true)
+    public boolean includeQcFailedReads = false;
+
     @ArgumentCollection
     public FlowBasedArgumentCollection fbargs = new FlowBasedArgumentCollection();
 
@@ -337,6 +344,11 @@ public void apply(final GATKRead read, final ReferenceContext referenceContext,
             return;
         }
 
+        // include qc-failed reads?
+        if ( ((read.getFlags() & VENDOR_QUALITY_CHECK_FLAG) != 0) && !includeQcFailedReads ) {
+            return;
+        }
+
         // flush qeues up to this read
         flushQueue(read, referenceContext);
 
diff --git a/src/test/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapperIntegrationTest.java b/src/test/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapperIntegrationTest.java
index 9958102b311..3663150b599 100644
--- a/src/test/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapperIntegrationTest.java
+++ b/src/test/java/org/broadinstitute/hellbender/tools/walkers/featuremapping/FlowFeatureMapperIntegrationTest.java
@@ -116,4 +116,35 @@ public void testSurroundingAllQuality() throws IOException {
         }
     }
 
+    @Test
+    public void testQCFailed() throws IOException {
+
+        final File outputDir = createTempDir("testFlowFeatureMapperTest");
+        final File expectedFile = new File(testDir + "/snv_feature_mapper_qc_failed_output.vcf");
+        final File outputFile = UPDATE_EXACT_MATCH_EXPECTED_OUTPUTS ? expectedFile : new File(outputDir + "/snv_feature_mapper_qc_failed_output.vcf");
+
+        final String[] args = new String[] {
+                "-R", largeFileTestDir + "/Homo_sapiens_assembly38.fasta.gz",
+                "-O", outputFile.getAbsolutePath(),
+                "-I", testDir + "/snv_feature_mapper_input.bam",
+                "--copy-attr", "tr",
+                "--limit-score", "100",
+                "--min-score", "0",
+                "--snv-identical-bases", "10",
+                "--debug-negatives", "false",
+                "--debug-read-name", "150451-BC94-0645901755",
+                "--include-qc-failed-reads", "true"
+        };
+
+        // run the tool
+        runCommandLine(args);  // no assert, just make sure we don't throw
+
+        // make sure we've generated the otuput file
+        Assert.assertTrue(outputFile.exists());
+
+        // walk the output and expected files, compare non-comment lines
+        if ( !UPDATE_EXACT_MATCH_EXPECTED_OUTPUTS ) {
+            IntegrationTestSpec.assertEqualTextFiles(outputFile, expectedFile, "#");
+        }
+    }
 }
diff --git a/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf b/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf
new file mode 100644
index 00000000000..0d5f50fceed
--- /dev/null
+++ b/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:405fb1bf62168e22c504e0e4ae0d9d6b9ab4f7ac32ff18dcd77f071618d964da
+size 5051930
diff --git a/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf.idx b/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf.idx
new file mode 100644
index 00000000000..d31b3036198
--- /dev/null
+++ b/src/test/resources/large/featureMapping/snv_feature_mapper_qc_failed_output.vcf.idx
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7aba6151d662570d8e4a25cbb4717c65009c2d217375a287667b1f47c2db9925
+size 120123