Added a '--prefer-mane-transcripts' mode that enforces MANE_Select tagged Gencode transcripts where possible (#9012)

* Added a '--prefer-mane-transcripts' mode that enforces MANE_Select tagged Gencode transcripts where possible

Author: jamesemery

src/main/java/org/broadinstitute/hellbender/tools/funcotator/BaseFuncotatorArgumentCollection.java

@@ -80,6 +80,14 @@ public abstract class BaseFuncotatorArgumentCollection implements Serializable {
     )
     public TranscriptSelectionMode transcriptSelectionMode = FuncotatorArgumentDefinitions.TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE;
 
+    @Advanced
+    @Argument(
+            fullName  = FuncotatorArgumentDefinitions.PREFER_MANE_TRANSCRIPT_MODE,
+            optional = true,
+            doc = "If this flag is set, Funcotator will prefer 'MANE_Plus_Clinical' followed by 'MANE_select' transcripts (including those not tagged 'basic') if one is present for a given variant. If neither tag is present it use the default behavior (only base transcripts)."
+    )
+    public boolean MANETranscriptMode = false;
+
     @Argument(
             fullName  = FuncotatorArgumentDefinitions.TRANSCRIPT_LIST_LONG_NAME,
             optional = true,

src/main/java/org/broadinstitute/hellbender/tools/funcotator/FuncotateSegments.java

@@ -147,7 +147,8 @@ public void onTraversalStart() {
                 new FlankSettings(0,0),
                 true,
                 funcotatorArgs.minNumBasesForValidSegment,
-                funcotatorArgs.spliceSiteWindow
+                funcotatorArgs.spliceSiteWindow,
+                funcotatorArgs.MANETranscriptMode
         ).stream()
          .filter(DataSourceFuncotationFactory::isSupportingSegmentFuncotation)
          .collect(Collectors.toList());

src/main/java/org/broadinstitute/hellbender/tools/funcotator/Funcotator.java

@@ -794,7 +794,8 @@ public void onTraversalStart() {
                 new FlankSettings(funcotatorArgs.fivePrimeFlankSize, funcotatorArgs.threePrimeFlankSize),
                 false,
                 funcotatorArgs.minNumBasesForValidSegment,
-                funcotatorArgs.spliceSiteWindow
+                funcotatorArgs.spliceSiteWindow,
+                funcotatorArgs.MANETranscriptMode
         );
 
         logger.info("Initializing Funcotator Engine...");

src/main/java/org/broadinstitute/hellbender/tools/funcotator/FuncotatorArgumentDefinitions.java

@@ -36,6 +36,8 @@ public class FuncotatorArgumentDefinitions {
     public static final String TRANSCRIPT_SELECTION_MODE_LONG_NAME = "transcript-selection-mode";
     public static final TranscriptSelectionMode TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE = TranscriptSelectionMode.CANONICAL;
 
+    public static final String PREFER_MANE_TRANSCRIPT_MODE = "prefer-mane-transcripts";
+
     /**
      * Do not give this a static default value or the integration tests will get hosed.
      */

src/main/java/org/broadinstitute/hellbender/tools/funcotator/dataSources/DataSourceUtils.java

@@ -329,6 +329,7 @@ private static boolean isValidDirectory(final Path p) {
      *                                                            ignored for those that don't.
      * @param minBasesForValidSegment The minimum number of bases for a segment to be considered valid.
      * @param spliceSiteWindowSize The number of bases on either side of a splice site for a variant to be a {@link org.broadinstitute.hellbender.tools.funcotator.dataSources.gencode.GencodeFuncotation.VariantClassification#SPLICE_SITE} variant.
+     * @param preferMANETranscriptsWhereApplicable If this is set, in {@link GencodeFuncotationFactory}, we will only emit MANE transcripts if any are availible for a given variant, otherwise behaves as normal.
      * @return A {@link List} of {@link DataSourceFuncotationFactory} given the data source metadata, overrides, and transcript reporting priority information.
      */
     public static List<DataSourceFuncotationFactory> createDataSourceFuncotationFactoriesForDataSources(final Map<Path, Properties> dataSourceMetaData,
@@ -340,7 +341,8 @@ public static List<DataSourceFuncotationFactory> createDataSourceFuncotationFact
                                                                                                         final FlankSettings flankSettings,
                                                                                                         final boolean doAttemptSegmentFuncotationForTranscriptDatasources,
                                                                                                         final int minBasesForValidSegment,
-                                                                                                        final int spliceSiteWindowSize) {
+                                                                                                        final int spliceSiteWindowSize,
+                                                                                                        final boolean preferMANETranscriptsWhereApplicable) {
         Utils.nonNull(dataSourceMetaData);
         Utils.nonNull(annotationOverridesMap);
         Utils.nonNull(transcriptSelectionMode);
@@ -379,7 +381,7 @@ public static List<DataSourceFuncotationFactory> createDataSourceFuncotationFact
                 case GENCODE:
                     featureInput = createAndRegisterFeatureInputs(path, properties, gatkToolInstance, lookaheadFeatureCachingInBp, GencodeGtfFeature.class, false);
                     funcotationFactory = DataSourceUtils.createGencodeDataSource(path, properties, annotationOverridesMap, transcriptSelectionMode,
-                            userTranscriptIdSet, featureInput, flankSettings, doAttemptSegmentFuncotationForTranscriptDatasources, minBasesForValidSegment, spliceSiteWindowSize);
+                            userTranscriptIdSet, featureInput, flankSettings, doAttemptSegmentFuncotationForTranscriptDatasources, minBasesForValidSegment, spliceSiteWindowSize, preferMANETranscriptsWhereApplicable);
                     break;
                 case VCF:
                     featureInput = createAndRegisterFeatureInputs(path, properties, gatkToolInstance, lookaheadFeatureCachingInBp, VariantContext.class, false);
@@ -596,7 +598,8 @@ private static GencodeFuncotationFactory createGencodeDataSource(final Path data
                                                                      final FlankSettings flankSettings,
                                                                      final boolean isSegmentFuncotationEnabled,
                                                                      final int minBasesForValidSegment,
-                                                                     final int spliceSiteWindowSize) {
+                                                                     final int spliceSiteWindowSize,
+                                                                     final boolean onlyUseMANETranscriptsWhenApplicable) {
         Utils.nonNull(dataSourceFile);
         Utils.nonNull(dataSourceProperties);
         Utils.nonNull(annotationOverridesMap);
@@ -626,7 +629,8 @@ private static GencodeFuncotationFactory createGencodeDataSource(final Path data
                 ncbiBuildVersion,
                 isSegmentFuncotationEnabled,
                 minBasesForValidSegment,
-                spliceSiteWindowSize
+                spliceSiteWindowSize,
+                onlyUseMANETranscriptsWhenApplicable
             );
     }
 

src/main/java/org/broadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java

@@ -242,6 +242,11 @@ public class GencodeFuncotationFactory extends DataSourceFuncotationFactory {
      */
     private boolean isSegmentFuncotationEnabled;
 
+    /**
+     * If this is true, only MANE transcripts will be used for funcotation creation when at least one is present.
+     */
+    private boolean preferMANETranscripts;
+
     //==================================================================================================================
     // Constructors:
 
@@ -354,7 +359,7 @@ public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFilePath,
 
         this(gencodeTranscriptFastaFilePath, version, name, transcriptSelectionMode, userRequestedTranscripts,
                 annotationOverrides, mainFeatureInput, flankSettings, isDataSourceB37, ncbiBuildVersion,
-                isSegmentFuncotationEnabled, minBasesForValidSegment, FuncotatorUtils.DEFAULT_SPLICE_SITE_WINDOW_SIZE);
+                isSegmentFuncotationEnabled, minBasesForValidSegment, FuncotatorUtils.DEFAULT_SPLICE_SITE_WINDOW_SIZE, false);
     }
 
     /**
@@ -385,7 +390,8 @@ public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFilePath,
                                      final String ncbiBuildVersion,
                                      final boolean isSegmentFuncotationEnabled,
                                      final int minBasesForValidSegment,
-                                     final int spliceSiteWindowSize) {
+                                     final int spliceSiteWindowSize,
+                                     final boolean preferMANETranscriptsWhereApplicable) {
 
         super(mainFeatureInput, minBasesForValidSegment);
 
@@ -429,6 +435,8 @@ public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFilePath,
 
         // Initialize overrides / defaults:
         initializeAnnotationOverrides( annotationOverrides );
+
+        this.preferMANETranscripts = preferMANETranscriptsWhereApplicable;
     }
 
     private Path localizeGencodeTranscriptFastaFile( final Path gencodeTranscriptFastaFilePath ) {
@@ -622,6 +630,28 @@ private static List<GencodeGtfGeneFeature> convertFeaturesToGencodeGtfGeneFeatur
                         .collect(Collectors.toList());
     }
 
+    /**
+     * If MANE_Plus_Clinical transcripts are avalible, only return them, followed by MANE_Select transcripts, followed by only the basic transcripts if none were MANE_Plus_Clinical or MANE_Select.
+     * @param transcripts of gencode transcripts to possibly filter
+     * @return
+     */
+    @VisibleForTesting
+    static List<GencodeGtfTranscriptFeature> retreiveMANESelectModeTranscriptsCriteria(final List<GencodeGtfTranscriptFeature> transcripts) {
+        final List<GencodeGtfTranscriptFeature> plusClincal = transcripts.stream()
+                .filter(g -> hasTag(g, MANE_PLUS_CLINICAL)).toList();
+        if (plusClincal.size() > 0) {
+            return plusClincal;
+        }
+
+        final List<GencodeGtfTranscriptFeature> maneSelectTranscripts = transcripts.stream()
+                .filter(g -> hasTag(g, MANE_SELECT)).toList();
+
+        if (maneSelectTranscripts.size() > 0) {
+            return maneSelectTranscripts;
+        }
+
+        return transcripts.stream().filter(GencodeFuncotationFactory::isBasic).collect(Collectors.toList());
+    }
 
     /**
      * {@inheritDoc}
@@ -853,16 +883,21 @@ static boolean isVariantInCodingRegion(final GencodeFuncotation.VariantClassific
      */
     private List<GencodeFuncotation> createFuncotationsHelper(final VariantContext variant, final Allele altAllele, final GencodeGtfGeneFeature gtfFeature, final ReferenceContext reference) {
 
-        final List<GencodeGtfTranscriptFeature> transcriptList;
+        List<GencodeGtfTranscriptFeature> transcriptList;
 
         // Only get basic transcripts if we're using data from Gencode:
         if ( gtfFeature.getGtfSourceFileType().equals(GencodeGtfCodec.GTF_FILE_TYPE_STRING) ) {
-            transcriptList = retrieveBasicTranscripts(gtfFeature);
-        }
-        else {
+            if (preferMANETranscripts) {
+                // Filter out the non-MANE_Select/Mane_Plus_Clinical transcripts if we're only using MANE transcripts:
+                transcriptList = retreiveMANESelectModeTranscriptsCriteria(gtfFeature.getTranscripts());
+            } else {
+                transcriptList = retrieveBasicTranscripts(gtfFeature);
+            }
+        } else {
             transcriptList = gtfFeature.getTranscripts();
         }
 
+
         return createFuncotationsHelper(variant, altAllele, reference, transcriptList);
     }
 
@@ -979,9 +1014,14 @@ static final GencodeFuncotation createDefaultFuncotationsOnProblemVariant( final
 
     private static boolean isBasic(final GencodeGtfTranscriptFeature transcript) {
         // Check if this transcript has the `basic` tag:
+        return hasTag(transcript, GencodeGTFFieldConstants.FeatureTag.BASIC);
+    }
+
+    private static boolean hasTag(final GencodeGtfTranscriptFeature transcript, final GencodeGTFFieldConstants.FeatureTag tag) {
+        // Check if this transcript has the given tag:
         return transcript.getOptionalFields().stream()
                 .filter( f -> f.getName().equals("tag") )
-                .filter( f -> f.getValue().equals(GencodeGTFFieldConstants.FeatureTag.BASIC.toString()) )
+                .filter( f -> f.getValue().equals(tag.toString()) )
                 .count() > 0;
     }
 

src/test/java/org/broadinstitute/hellbender/tools/funcotator/FuncotatorEngineUnitTest.java

@@ -65,7 +65,8 @@ public void testGetFuncotationFactoriesAndCreateFuncotationMapForVariant(final F
                                 new FlankSettings(0, 0),
                                 false,
                                 FuncotatorUtils.DEFAULT_MIN_NUM_BASES_FOR_VALID_SEGMENT,
-                                FuncotatorUtils.DEFAULT_SPLICE_SITE_WINDOW_SIZE)
+                                FuncotatorUtils.DEFAULT_SPLICE_SITE_WINDOW_SIZE,
+                                false)
                 );
 
         for (int i = 0; i < entireVcf.getRight().size(); i++) {