Modified HaplotypeBasedVariantRecaller to support non-flow reads (#8896)

broadinstitute · Jul 24, 2024 · 747df1a · 747df1a
1 parent 59c9c1b
commit 747df1a
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 15 deletions.
diff --git a/...oadinstitute/hellbender/tools/walkers/variantrecalling/HaplotypeBasedVariantRecaller.java b/...oadinstitute/hellbender/tools/walkers/variantrecalling/HaplotypeBasedVariantRecaller.java
@@ -135,11 +135,11 @@ public void traverse() {
                     }
 
                     // get reads overlapping haplotypes
-                    final Map<SamReader, Collection<FlowBasedRead>> readsByReader = readsReader.getReads(haplotypeSpan, vcLoc);
+                    final Map<SamReader, Collection<GATKRead>> readsByReader = readsReader.getReads(haplotypeSpan, vcLoc);
                     final List<VariantContext>      variants = new LinkedList<>(Arrays.asList(vc));
                     if ( logger.isDebugEnabled() ) {
                         int readCount = 0;
-                        for ( Collection<FlowBasedRead> reads : readsByReader.values() )
+                        for ( Collection<GATKRead> reads : readsByReader.values() )
                             readCount += reads.size();
                         logger.debug(String.format("vcLoc %s, haplotypeSpan: %s, %d haplotypes, %d reads",
                                 vcLoc.toString(), haplotypeSpan.toString(), processedHaplotypes.size(), readCount, variants.size()));
@@ -150,16 +150,16 @@ public void traverse() {
                     final List<Map<Integer, AlleleLikelihoods<GATKRead, Allele>>> genotypeLikelihoodsList = new LinkedList<>();
                     final List<AssemblyResultSet>                                 assemblyResultList = new LinkedList<>();
                     final List<SAMFileHeader>                                     readsHeaderList = new LinkedList<>();
-                    for ( Map.Entry<SamReader, Collection<FlowBasedRead>> entry : readsByReader.entrySet() ) {
+                    for ( Map.Entry<SamReader, Collection<GATKRead>> entry : readsByReader.entrySet() ) {
                         final AssemblyResultSet assemblyResult = new AssemblyResultSet();
                         processedHaplotypes.forEach(haplotype -> assemblyResult.add(haplotype));
 
                         final Map<String, List<GATKRead>> perSampleReadList = new LinkedHashMap<>();
                         final SamReader                   samReader = entry.getKey();
-                        final Collection<FlowBasedRead>   reads = entry.getValue();
+                        final Collection<GATKRead>   reads = entry.getValue();
 
                         List<GATKRead> gtakReads = new LinkedList<>();
-                        reads.forEach(flowBasedRead -> gtakReads.add(flowBasedRead));
+                        reads.forEach(read -> gtakReads.add(read));
                         perSampleReadList.put(sampleNames[0], gtakReads);
                         AssemblyRegion regionForGenotyping = new AssemblyRegion(haplotypeSpan, 0, samReader.getFileHeader());
                         assemblyResult.setPaddedReferenceLoc(haplotypeSpan);

diff --git a/...java/org/broadinstitute/hellbender/tools/walkers/variantrecalling/TrimmedReadsReader.java b/...java/org/broadinstitute/hellbender/tools/walkers/variantrecalling/TrimmedReadsReader.java
@@ -26,8 +26,6 @@ public class TrimmedReadsReader {
 
     private final List<SamReader>         samReaders = new LinkedList<>();
     private CountingReadFilter            readFilter;
-    private final Map<String, Integer>    readGroupMaxClass = new LinkedHashMap<>();
-    private final Map<String, String>     readGroupFlowOrder = new LinkedHashMap<>();
     private final FlowBasedArgumentCollection fbArgs = new FlowBasedArgumentCollection();
 
     public TrimmedReadsReader(final List<Path> readsFiles, final Path referencePath, final int cloudPrefetchBuffer) {
@@ -44,11 +42,12 @@ public SAMSequenceDictionary getSamSequenceDictionary(final SamReader samReader)
         return ((samReader != null) ? samReader : samReaders.get(0)).getFileHeader().getSequenceDictionary();
     }
 
-    public Map<SamReader, Collection<FlowBasedRead>>  getReads(final Locatable span, final Locatable vcLoc) {
 
-        final Map<SamReader, Collection<FlowBasedRead>>   readsByReader = new LinkedHashMap<>();
+    public Map<SamReader, Collection<GATKRead>>  getReads(final Locatable span, final Locatable vcLoc) {
+
+        final Map<SamReader, Collection<GATKRead>>   readsByReader = new LinkedHashMap<>();
         for ( SamReader samReader : samReaders ) {
-            final List<FlowBasedRead>     reads = new LinkedList<>();
+            final List<GATKRead>     reads = new LinkedList<>();
             final SAMRecordIterator iter = samReader.query(span.getContig(), span.getStart(), span.getEnd(), false);
             while (iter.hasNext()) {
 
@@ -72,7 +71,10 @@ public Map<SamReader, Collection<FlowBasedRead>>  getReads(final Locatable span,
                 gatkRead = ReadClipper.hardClipToRegion(gatkRead, span.getStart(), span.getEnd());
                 if (gatkRead.isUnmapped() || gatkRead.getCigar().isEmpty())
                     continue;
-
+                if (!FlowBasedReadUtils.isFlowPlatform(samReader.getFileHeader(), gatkRead)){
+                    reads.add(gatkRead);
+                    continue;
+                }
                 // convert to a flow based read
                 FlowBasedReadUtils.ReadGroupInfo rgInfo = FlowBasedReadUtils.getReadGroupInfo(samReader.getFileHeader(), gatkRead);
                 final FlowBasedRead fbr = new FlowBasedRead(gatkRead, rgInfo.flowOrder, rgInfo.maxClass, fbArgs);

diff --git a/.../broadinstitute/hellbender/tools/walkers/variantrecalling/TrimmedReadsReaderUnitTest.java b/.../broadinstitute/hellbender/tools/walkers/variantrecalling/TrimmedReadsReaderUnitTest.java
@@ -5,6 +5,7 @@
 import org.broadinstitute.hellbender.GATKBaseTest;
 import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.read.FlowBasedRead;
+import org.broadinstitute.hellbender.utils.read.GATKRead;
 import org.testng.Assert;
 import org.testng.annotations.DataProvider;
 import org.testng.annotations.Test;
@@ -68,10 +69,10 @@ public void testBasic(final BamSource bamSources[], final Locatable span, final
         Assert.assertNotNull(reader.getHeader(null));
 
         // reads
-        Map<SamReader, Collection<FlowBasedRead>> reads = reader.getReads(span, vcLoc);
+        Map<SamReader, Collection<GATKRead>> reads = reader.getReads(span, vcLoc);
         Assert.assertEquals(reads.size(), bamSources.length);
         int             bamSourceIndex = 0;
-        for ( Map.Entry<SamReader, Collection<FlowBasedRead>> entry : reads.entrySet() ) {
+        for ( Map.Entry<SamReader, Collection<GATKRead>> entry : reads.entrySet() ) {
 
             final BamSource       bamSource = bamSources[bamSourceIndex++];
 
@@ -83,8 +84,8 @@ public void testBasic(final BamSource bamSources[], final Locatable span, final
             Assert.assertEquals(entry.getValue().size(), bamSource.readCount);
 
             // verify first and last
-            FlowBasedRead       firstRead = entry.getValue().iterator().next();
-            FlowBasedRead       lastRead = entry.getValue().stream().reduce((prev, next) -> next).orElse(null);
+            FlowBasedRead       firstRead = (FlowBasedRead) entry.getValue().iterator().next();
+            FlowBasedRead       lastRead = (FlowBasedRead) entry.getValue().stream().reduce((prev, next) -> next).orElse(null);
             Assert.assertEquals(firstRead.getName(), bamSource.firstReadName);
             Assert.assertEquals(lastRead.getName(), bamSource.lastReadName);
         }