JAL-3076 refactor for more efficient scan of 'gene' features

[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 cdcfa96..36b19e2 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -100,6 +100,12 @@ 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
@@ -149,7 +155,11 @@ public class EnsemblGene extends EnsemblSeqProxy
       
       if (geneAlignment.getHeight() == 1)
       {
+        // ensure id has 'correct' case for the Ensembl identifier
+        geneId = geneAlignment.getSequenceAt(0).getName();
+
         findGeneLoci(geneAlignment.getSequenceAt(0), geneId);
+
         getTranscripts(geneAlignment, geneId);
       }
       if (al == null)
@@ -235,7 +245,7 @@ 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()))
     {
@@ -370,7 +380,7 @@ 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)
     {
@@ -412,7 +422,7 @@ 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());
@@ -498,7 +508,7 @@ 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;
 
@@ -510,7 +520,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     for (SequenceFeature sf : sfs)
     {
       String parent = (String) sf.getValue(PARENT);
-      if (parentIdentifier.equals(parent))
+      if (parentIdentifier.equalsIgnoreCase(parent))
       {
         transcriptFeatures.add(sf);
       }
@@ -538,22 +548,27 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Answers true for a feature of type 'gene' (or a sub-type of gene in the
-   * Sequence Ontology), whose ID is the accession we are retrieving
+   * Answers a list of sequence features (if any) whose type is 'gene' (or a
+   * subtype of gene in the Sequence Ontology), and whose ID is the accession we
+   * are retrieving
    */
   @Override
-  protected boolean identifiesSequence(SequenceFeature sf, String accId)
+  protected List<SequenceFeature> getIdentifyingFeatures(SequenceI seq,
+          String accId)
   {
-    if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
-            SequenceOntologyI.GENE))
+    List<SequenceFeature> result = new ArrayList<>();
+    List<SequenceFeature> sfs = seq.getFeatures()
+            .getFeaturesByOntology(SequenceOntologyI.GENE);
+    for (SequenceFeature sf : sfs)
     {
-      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;
+        result.add(sf);
       }
     }
-    return false;
+    return result;
   }
 
   /**
@@ -575,7 +590,7 @@ 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;
       }
@@ -584,17 +599,6 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Answers false. This allows an optimisation - a single 'gene' feature is all
-   * that is needed to identify the positions of the gene on the genomic
-   * sequence.
-   */
-  @Override
-  protected boolean isSpliceable()
-  {
-    return false;
-  }
-
-  /**
    * Override to do nothing as Ensembl doesn't return a protein sequence for a
    * gene identifier
    */