JAL-2679 update query id to actual case of retrieved gene

[jalview.git] / src / jalview / ext / ensembl / EnsemblGene.java

diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java
index f46cb1b..0d5fc26 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -98,6 +98,12 @@ public class EnsemblGene extends EnsemblSeqProxy
     return EnsemblSeqType.GENOMIC;
   }
 
+  @Override
+  protected String getObjectType()
+  {
+    return OBJECT_TYPE_GENE;
+  }
+
   /**
    * Returns an alignment containing the gene(s) for the given gene or
    * transcript identifier, or external identifier (e.g. Uniprot id). If given a
@@ -146,6 +152,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       }
       if (geneAlignment.getHeight() == 1)
       {
+        // ensure id has 'correct' case for the Ensembl identifier
+        geneId = geneAlignment.getSequenceAt(0).getName();
         getTranscripts(geneAlignment, geneId);
       }
       if (al == null)
@@ -170,7 +178,7 @@ public class EnsemblGene extends EnsemblSeqProxy
    */
   List<String> getGeneIds(String accessions)
   {
-    List<String> geneIds = new ArrayList<String>();
+    List<String> geneIds = new ArrayList<>();
 
     for (String acc : accessions.split(getAccessionSeparator()))
     {
@@ -305,7 +313,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     int transcriptLength = 0;
     final char[] geneChars = gene.getSequence();
     int offset = gene.getStart(); // to convert to 0-based positions
-    List<int[]> mappedFrom = new ArrayList<int[]>();
+    List<int[]> mappedFrom = new ArrayList<>();
 
     for (SequenceFeature sf : splices)
     {
@@ -347,7 +355,7 @@ public class EnsemblGene extends EnsemblSeqProxy
      * transfer features to the new sequence; we use EnsemblCdna to do this,
      * to filter out unwanted features types (see method retainFeature)
      */
-    List<int[]> mapTo = new ArrayList<int[]>();
+    List<int[]> mapTo = new ArrayList<>();
     mapTo.add(new int[] { 1, transcriptLength });
     MapList mapping = new MapList(mappedFrom, mapTo, 1, 1);
     EnsemblCdna cdna = new EnsemblCdna(getDomain());
@@ -381,6 +389,12 @@ public class EnsemblGene extends EnsemblSeqProxy
   /**
    * Returns a list of the transcript features on the sequence whose Parent is
    * the gene for the accession id.
+   * <p>
+   * Transcript features are those of type "transcript", or any of its sub-types
+   * in the Sequence Ontology e.g. "mRNA", "processed_transcript". We also
+   * include "NMD_transcript_variant", because this type behaves like a
+   * transcript identifier in Ensembl, although strictly speaking it is not in
+   * the SO.
    * 
    * @param accId
    * @param geneSequence
@@ -389,22 +403,21 @@ public class EnsemblGene extends EnsemblSeqProxy
   protected List<SequenceFeature> getTranscriptFeatures(String accId,
           SequenceI geneSequence)
   {
-    List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
+    List<SequenceFeature> transcriptFeatures = new ArrayList<>();
 
     String parentIdentifier = GENE_PREFIX + accId;
-    // todo optimise here by transcript type!
+
     List<SequenceFeature> sfs = geneSequence.getFeatures()
-            .getPositionalFeatures();
+            .getFeaturesByOntology(SequenceOntologyI.TRANSCRIPT);
+    sfs.addAll(geneSequence.getFeatures().getPositionalFeatures(
+            SequenceOntologyI.NMD_TRANSCRIPT_VARIANT));
 
     for (SequenceFeature sf : sfs)
     {
-      if (isTranscript(sf.getType()))
+      String parent = (String) sf.getValue(PARENT);
+      if (parentIdentifier.equalsIgnoreCase(parent))
       {
-        String parent = (String) sf.getValue(PARENT);
-        if (parentIdentifier.equals(parent))
-        {
-          transcriptFeatures.add(sf);
-        }
+        transcriptFeatures.add(sf);
       }
     }
 
@@ -439,8 +452,9 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
             SequenceOntologyI.GENE))
     {
-      String id = (String) sf.getValue(ID);
-      if ((GENE_PREFIX + accId).equals(id))
+      // NB features as gff use 'ID'; rest services return as 'id'
+      String id = (String) sf.getValue("ID");
+      if ((GENE_PREFIX + accId).equalsIgnoreCase(id))
       {
         return true;
       }
@@ -467,7 +481,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (isTranscript(type))
     {
       String parent = (String) sf.getValue(PARENT);
-      if (!(GENE_PREFIX + accessionId).equals(parent))
+      if (!(GENE_PREFIX + accessionId).equalsIgnoreCase(parent))
       {
         return false;
       }