Fixed support for newer Gencode GTF versions by making the GencodeGTF…

…Field parsing more permissive (#8351)
broadinstitute · Jun 28, 2023 · 01e45a2 · 01e45a2
1 parent daeb3e2
commit 01e45a2
Show file tree

Hide file tree

Showing 17 changed files with 2,886 additions and 3,012 deletions.
diff --git a/src/main/java/org/broadinstitute/hellbender/tools/funcotator/TranscriptSelectionMode.java b/src/main/java/org/broadinstitute/hellbender/tools/funcotator/TranscriptSelectionMode.java
@@ -1,6 +1,7 @@
 package org.broadinstitute.hellbender.tools.funcotator;
 
 import org.broadinstitute.hellbender.tools.funcotator.dataSources.gencode.GencodeFuncotation;
+import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGTFFieldConstants;
 import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGtfFeature;
 
 import java.util.Comparator;
@@ -246,8 +247,8 @@ public ComparatorByProteinCodingStatus(){}
         @Override
         public int compare( final GencodeFuncotation a, final GencodeFuncotation b ) {
             // Is it protein coding?
-            final boolean isAProteinCoding = GencodeGtfFeature.KnownGeneBiotype.PROTEIN_CODING.toString().equals(a.getGeneTranscriptType());
-            final boolean isBProteinCoding = GencodeGtfFeature.KnownGeneBiotype.PROTEIN_CODING.toString().equals(b.getGeneTranscriptType());
+            final boolean isAProteinCoding = GencodeGTFFieldConstants.KnownGeneBiotype.PROTEIN_CODING.toString().equals(a.getGeneTranscriptType());
+            final boolean isBProteinCoding = GencodeGTFFieldConstants.KnownGeneBiotype.PROTEIN_CODING.toString().equals(b.getGeneTranscriptType());
             if ( isAProteinCoding != isBProteinCoding ) {
                 if ( isAProteinCoding ) {
                     return -1;

diff --git a/...rg/broadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotation.java b/...rg/broadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotation.java
@@ -7,6 +7,7 @@
 import org.broadinstitute.hellbender.tools.funcotator.Funcotation;
 import org.broadinstitute.hellbender.tools.funcotator.metadata.FuncotationMetadata;
 import org.broadinstitute.hellbender.tools.funcotator.vcfOutput.VcfOutputRenderer;
+import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGTFFieldConstants;
 import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGtfGeneFeature;
 
 import java.util.Arrays;
@@ -74,7 +75,7 @@ public class GencodeFuncotation implements Funcotation {
 
     // These are included because they help determine the transcript selection
     private Integer                              locusLevel;
-    private GencodeGtfGeneFeature.FeatureTag     apprisRank;
+    private GencodeGTFFieldConstants.FeatureTag apprisRank;
     private Integer                              transcriptLength;
     private String                               version;
     private String                               geneTranscriptType;
@@ -378,7 +379,7 @@ public boolean equals(final Object o) {
         if (transcriptLength != null ? !transcriptLength.equals(that.transcriptLength) : that.transcriptLength != null)
             return false;
         if (version != null ? !version.equals(that.version) : that.version != null) return false;
-        if (geneTranscriptType != that.geneTranscriptType) return false;
+        if (geneTranscriptType != that.geneTranscriptType) return false; //TODO this is a problem string equality comparison.... it breaks tests to fix it though...
         if (hugoSymbolSerializedOverride != null ? !hugoSymbolSerializedOverride.equals(that.hugoSymbolSerializedOverride) : that.hugoSymbolSerializedOverride != null)
             return false;
         if (ncbiBuildSerializedOverride != null ? !ncbiBuildSerializedOverride.equals(that.ncbiBuildSerializedOverride) : that.ncbiBuildSerializedOverride != null)
@@ -660,11 +661,11 @@ public void setLocusLevel(final Integer locusLevel) {
         this.locusLevel = locusLevel;
     }
 
-    public GencodeGtfGeneFeature.FeatureTag getApprisRank() {
+    public GencodeGTFFieldConstants.FeatureTag getApprisRank() {
         return apprisRank;
     }
 
-    public void setApprisRank(final GencodeGtfGeneFeature.FeatureTag apprisRank) {
+    public void setApprisRank(final GencodeGTFFieldConstants.FeatureTag apprisRank) {
         this.apprisRank = apprisRank;
     }
 

diff --git a/...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationBuilder.java b/...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationBuilder.java
@@ -5,6 +5,7 @@
 import htsjdk.variant.variantcontext.Allele;
 import org.broadinstitute.hellbender.exceptions.GATKException;
 import org.broadinstitute.hellbender.tools.funcotator.metadata.FuncotationMetadataUtils;
+import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGTFFieldConstants;
 import org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGtfGeneFeature;
 
 import java.util.ArrayList;
@@ -260,10 +261,10 @@ public GencodeFuncotationBuilder setLocusLevel( final Integer locusLevel ) {
 
     /**
      * Set the Appris Rank in the {@link GencodeFuncotation}.
-     * @param apprisRank The {@link GencodeGtfGeneFeature.FeatureTag} containing the Appris Rank for the {@link GencodeFuncotation}.
+     * @param apprisRank The {@link GencodeGTFFieldConstants.FeatureTag} containing the Appris Rank for the {@link GencodeFuncotation}.
      * @return {@code this} {@link GencodeFuncotationBuilder}
      */
-    public GencodeFuncotationBuilder setApprisRank( final GencodeGtfGeneFeature.FeatureTag apprisRank ) {
+    public GencodeFuncotationBuilder setApprisRank( final GencodeGTFFieldConstants.FeatureTag apprisRank ) {
         gencodeFuncotation.setApprisRank( apprisRank );
         return this;
     }

diff --git a/...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java b/...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java
@@ -46,7 +46,7 @@
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
 
-import static org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGtfFeature.FeatureTag.*;
+import static org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGTFFieldConstants.FeatureTag.*;
 
 /**
  * A factory to create {@link GencodeFuncotation}s.
@@ -111,24 +111,22 @@ public class GencodeFuncotationFactory extends DataSourceFuncotationFactory {
     /**
      * List of valid Appris Ranks used for sorting funcotations to get the "best" one.z
      */
-    private static final HashSet<GencodeGtfGeneFeature.FeatureTag> apprisRanks = new HashSet<>(
-            Arrays.asList(
-                APPRIS_PRINCIPAL,
-                APPRIS_PRINCIPAL_1,
-                APPRIS_PRINCIPAL_2,
-                APPRIS_PRINCIPAL_3,
-                APPRIS_PRINCIPAL_4,
-                APPRIS_PRINCIPAL_5,
-                APPRIS_ALTERNATIVE_1,
-                APPRIS_ALTERNATIVE_2,
-                APPRIS_CANDIDATE_HIGHEST_SCORE,
-                APPRIS_CANDIDATE_LONGEST_CCDS,
-                APPRIS_CANDIDATE_CCDS,
-                APPRIS_CANDIDATE_LONGEST_SEQ,
-                APPRIS_CANDIDATE_LONGEST,
-                APPRIS_CANDIDATE
-            )
-    );
+    private static final LinkedHashMap<String, GencodeGTFFieldConstants.FeatureTag> apprisRanks = new LinkedHashMap<>() {{
+            put(APPRIS_PRINCIPAL.toString(), APPRIS_PRINCIPAL);
+            put(APPRIS_PRINCIPAL_1.toString(), APPRIS_PRINCIPAL_1);
+            put(APPRIS_PRINCIPAL_2.toString(), APPRIS_PRINCIPAL_2);
+            put(APPRIS_PRINCIPAL_3.toString(), APPRIS_PRINCIPAL_3);
+            put(APPRIS_PRINCIPAL_4.toString(), APPRIS_PRINCIPAL_4);
+            put(APPRIS_PRINCIPAL_5.toString(), APPRIS_PRINCIPAL_5);
+            put(APPRIS_ALTERNATIVE_1.toString(), APPRIS_ALTERNATIVE_1);
+            put(APPRIS_ALTERNATIVE_2.toString(), APPRIS_ALTERNATIVE_2);
+            put(APPRIS_CANDIDATE_HIGHEST_SCORE.toString(), APPRIS_CANDIDATE_HIGHEST_SCORE);
+            put(APPRIS_CANDIDATE_LONGEST_CCDS.toString(), APPRIS_CANDIDATE_LONGEST_CCDS);
+            put(APPRIS_CANDIDATE_CCDS.toString(), APPRIS_CANDIDATE_CCDS);
+            put(APPRIS_CANDIDATE_LONGEST_SEQ.toString(), APPRIS_CANDIDATE_LONGEST_SEQ);
+            put(APPRIS_CANDIDATE_LONGEST.toString(), APPRIS_CANDIDATE_LONGEST);
+            put(APPRIS_CANDIDATE.toString(), APPRIS_CANDIDATE);
+        }};
 
     /**
      * The set of {@link GencodeFuncotation.VariantClassification} types that are valid for coding regions.
@@ -946,8 +944,7 @@ private static boolean isBasic(final GencodeGtfTranscriptFeature transcript) {
         // Check if this transcript has the `basic` tag:
         return transcript.getOptionalFields().stream()
                 .filter( f -> f.getName().equals("tag") )
-                .filter( f -> f.getValue() instanceof GencodeGtfFeature.FeatureTag )
-                .filter( f -> f.getValue().equals(GencodeGtfFeature.FeatureTag.BASIC) )
+                .filter( f -> f.getValue().equals(GencodeGTFFieldConstants.FeatureTag.BASIC.toString()) )
                 .count() > 0;
     }
 
@@ -1079,7 +1076,7 @@ private GencodeFuncotation createExonFuncotation(final VariantContext variant,
 
         // Before we get started, check to see if this is a non-protein-coding feature.
         // If it is, we must handle it differently:
-        if ( GencodeGtfFeature.KnownGeneBiotype.PROTEIN_CODING.toString().equals(transcript.getGeneType()) ) {
+        if ( GencodeGTFFieldConstants.KnownGeneBiotype.PROTEIN_CODING.toString().equals(transcript.getGeneType()) ) {
             return createCodingRegionFuncotationForProteinCodingFeature(variant, altAllele, reference, transcript, exon);
         }
         else {
@@ -1700,7 +1697,7 @@ private GencodeFuncotation createIntronFuncotation(final VariantContext variant,
         gencodeFuncotationBuilder.setReferenceContext(referenceBases.getBaseString(Strand.POSITIVE));
 
         // Set the VariantClassification:
-        if ( GencodeGtfFeature.KnownGeneBiotype.PROTEIN_CODING.toString().equals(transcript.getGeneType()) ) {
+        if ( GencodeGTFFieldConstants.KnownGeneBiotype.PROTEIN_CODING.toString().equals(transcript.getGeneType()) ) {
             gencodeFuncotationBuilder.setVariantClassification(GencodeFuncotation.VariantClassification.INTRON);
         }
         else {
@@ -2708,19 +2705,18 @@ else if (altAllele.length() < refAllele.length()) {
 
     /**
      * Get the Appris Rank from the given {@link GencodeGtfGeneFeature}.
-     * Appris ranks are specified as annotations using {@link org.broadinstitute.hellbender.utils.codecs.gtf.GencodeGtfFeature.FeatureTag}s.
+     * Appris ranks are specified as annotations using {@link GencodeGTFFieldConstants.FeatureTag}s.
      * @param gtfFeature The {@link GencodeGtfTranscriptFeature} from which to get the Appris Rank.
      * @return The highest Appris Rank found in the given {@code gtfFeature}; if no Appris Rank exists, {@code null}.
      */
     @VisibleForTesting
-    static GencodeGtfFeature.FeatureTag getApprisRank( final GencodeGtfTranscriptFeature gtfFeature ) {
+    static GencodeGTFFieldConstants.FeatureTag getApprisRank(final GencodeGtfTranscriptFeature gtfFeature ) {
 
-        // Get our appris tag(s) if it/they exist(s):
-        final List<GencodeGtfFeature.FeatureTag> gtfApprisTags = gtfFeature.getOptionalFields().stream()
+        // Get the Appris Rank tags and convert them to Sortable Enums:
+        final List<GencodeGTFFieldConstants.FeatureTag> gtfApprisTags = gtfFeature.getOptionalFields().stream()
                 .filter( f -> f.getName().equals("tag") )
-                .filter( f -> f.getValue() instanceof GencodeGtfFeature.FeatureTag )
-                .filter( f -> apprisRanks.contains( f.getValue() ) )
-                .map( f -> (GencodeGtfFeature.FeatureTag)f.getValue() ).collect(Collectors.toList());
+                .filter( f -> apprisRanks.containsKey( f.getValue() ) )
+                .map( f -> apprisRanks.get(f.getValue()) ).collect(Collectors.toList());
 
         if ( gtfApprisTags.isEmpty() ) {
             return null;
@@ -2737,16 +2733,16 @@ else if ( gtfApprisTags.size() == 1 ) {
 
     /**
      * Converts a given GeneTranscriptType {@link String} to a {@link org.broadinstitute.hellbender.tools.funcotator.dataSources.gencode.GencodeFuncotation.VariantClassification}.
-     * Assumes the given {@code type} is not {@link GencodeGtfFeature.KnownGeneBiotype#PROTEIN_CODING}.
+     * Assumes the given {@code type} is not {@link GencodeGTFFieldConstants.KnownGeneBiotype#PROTEIN_CODING}.
      * If no type can be assessed, returns {@code null}.
      * @param type A {@link String} representing a GeneTranscriptType to convert to a {@link org.broadinstitute.hellbender.tools.funcotator.dataSources.gencode.GencodeFuncotation.VariantClassification}.
      * @return A {@link org.broadinstitute.hellbender.tools.funcotator.dataSources.gencode.GencodeFuncotation.VariantClassification} representing the given GeneTranscriptType {@link String}, or {@code null}.
      */
     private static GencodeFuncotation.VariantClassification convertGeneTranscriptTypeToVariantClassification (final String type ) {
 
         //TODO: This all needs to be fixed so there is a 1:1 mapping of GencodeGtfFeature.KnownGeneBiotype->VariantClassification - Issue #4405
-        if (GencodeGtfFeature.KnownGeneBiotype.LINCRNA.toString().equals(type) ||
-                GencodeGtfFeature.KnownGeneBiotype.MACRO_LNCRNA.toString().equals(type)) {
+        if (GencodeGTFFieldConstants.KnownGeneBiotype.LINCRNA.toString().equals(type) ||
+                GencodeGTFFieldConstants.KnownGeneBiotype.MACRO_LNCRNA.toString().equals(type)) {
 			return GencodeFuncotation.VariantClassification.LINCRNA;
 		}
         return GencodeFuncotation.VariantClassification.RNA;