JAL-2777 test for the whitespace chain

[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 edeeedd..ad01324 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -26,6 +26,7 @@ import jalview.datamodel.AlignmentI;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
+import jalview.datamodel.features.SequenceFeatures;
 import jalview.io.gff.SequenceOntologyFactory;
 import jalview.io.gff.SequenceOntologyI;
 import jalview.schemes.FeatureColour;
@@ -283,22 +284,20 @@ public class EnsemblGene extends EnsemblSeqProxy
    */
   protected void clearGeneFeatures(SequenceI gene)
   {
-    SequenceFeature[] sfs = gene.getSequenceFeatures();
-    if (sfs != null)
+    /*
+     * Note we include NMD_transcript_variant here because it behaves like 
+     * 'transcript' in Ensembl, although strictly speaking it is not 
+     * (it is a sub-type of sequence_variant)    
+     */
+    String[] soTerms = new String[] {
+        SequenceOntologyI.NMD_TRANSCRIPT_VARIANT,
+        SequenceOntologyI.TRANSCRIPT, SequenceOntologyI.EXON,
+        SequenceOntologyI.CDS };
+    List<SequenceFeature> sfs = gene.getFeatures().getFeaturesByOntology(
+            soTerms);
+    for (SequenceFeature sf : sfs)
     {
-      SequenceOntologyI so = SequenceOntologyFactory.getInstance();
-      List<SequenceFeature> filtered = new ArrayList<SequenceFeature>();
-      for (SequenceFeature sf : sfs)
-      {
-        String type = sf.getType();
-        if (!isTranscript(type) && !so.isA(type, SequenceOntologyI.EXON)
-                && !so.isA(type, SequenceOntologyI.CDS))
-        {
-          filtered.add(sf);
-        }
-      }
-      gene.setSequenceFeatures(
-              filtered.toArray(new SequenceFeature[filtered.size()]));
+      gene.deleteFeature(sf);
     }
   }
 
@@ -348,6 +347,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     {
       splices = findFeatures(gene, SequenceOntologyI.CDS, parentId);
     }
+    SequenceFeatures.sortFeatures(splices, true);
 
     int transcriptLength = 0;
     final char[] geneChars = gene.getSequence();
@@ -398,7 +398,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     mapTo.add(new int[] { 1, transcriptLength });
     MapList mapping = new MapList(mappedFrom, mapTo, 1, 1);
     EnsemblCdna cdna = new EnsemblCdna(getDomain());
-    cdna.transferFeatures(gene.getSequenceFeatures(),
+    cdna.transferFeatures(gene.getFeatures().getPositionalFeatures(),
             transcript.getDatasetSequence(), mapping, parentId);
 
     /*
@@ -428,6 +428,12 @@ 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
@@ -439,20 +445,18 @@ public class EnsemblGene extends EnsemblSeqProxy
     List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
 
     String parentIdentifier = GENE_PREFIX + accId;
-    SequenceFeature[] sfs = geneSequence.getSequenceFeatures();
 
-    if (sfs != null)
+    List<SequenceFeature> sfs = geneSequence.getFeatures()
+            .getFeaturesByOntology(SequenceOntologyI.TRANSCRIPT);
+    sfs.addAll(geneSequence.getFeatures().getPositionalFeatures(
+            SequenceOntologyI.NMD_TRANSCRIPT_VARIANT));
+
+    for (SequenceFeature sf : sfs)
     {
-      for (SequenceFeature sf : sfs)
+      String parent = (String) sf.getValue(PARENT);
+      if (parentIdentifier.equals(parent))
       {
-        if (isTranscript(sf.getType()))
-        {
-          String parent = (String) sf.getValue(PARENT);
-          if (parentIdentifier.equals(parent))
-          {
-            transcriptFeatures.add(sf);
-          }
-        }
+        transcriptFeatures.add(sf);
       }
     }