JAL-2446 merged to spike branch

[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 24e3e95..915fa0a 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;
@@ -190,7 +191,22 @@ public class EnsemblGene extends EnsemblSeqProxy
           geneIds.add(geneId);
         }
       }
+      else if (isProteinIdentifier(acc))
+      {
+        String tscriptId = new EnsemblLookup(getDomain()).getParent(acc);
+        if (tscriptId != null)
+        {
+          String geneId = new EnsemblLookup(getDomain())
+                  .getParent(tscriptId);
 
+          if (geneId != null && !geneIds.contains(geneId))
+          {
+            geneIds.add(geneId);
+          }
+        }
+        // NOTE - acc is lost if it resembles an ENS.+ ID but isn't actually
+        // resolving to one... e.g. ENSMICP00000009241
+      }
       /*
        * if given a gene or other external name, lookup and fetch 
        * the corresponding gene for all model organisms 
@@ -267,22 +283,20 @@ 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);
     }
   }
 
@@ -332,6 +346,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     {
       splices = findFeatures(gene, SequenceOntologyI.CDS, parentId);
     }
+    SequenceFeatures.sortFeatures(splices, true);
 
     int transcriptLength = 0;
     final char[] geneChars = gene.getSequence();
@@ -381,7 +396,7 @@ 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);
 
     /*
@@ -422,19 +437,18 @@ public class EnsemblGene extends EnsemblSeqProxy
     List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
 
     String parentIdentifier = GENE_PREFIX + accId;
-    SequenceFeature[] sfs = geneSequence.getSequenceFeatures();
+    // todo optimise here by transcript type!
+    List<SequenceFeature> sfs = geneSequence.getFeatures()
+            .getPositionalFeatures();
 
-    if (sfs != null)
+    for (SequenceFeature sf : sfs)
     {
-      for (SequenceFeature sf : sfs)
+      if (isTranscript(sf.getType()))
       {
-        if (isTranscript(sf.getType()))
+        String parent = (String) sf.getValue(PARENT);
+        if (parentIdentifier.equals(parent))
         {
-          String parent = (String) sf.getValue(PARENT);
-          if (parentIdentifier.equals(parent))
-          {
-            transcriptFeatures.add(sf);
-          }
+          transcriptFeatures.add(sf);
         }
       }
     }