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 2d4d61a..ad01324 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -110,7 +110,8 @@ public class EnsemblGene extends EnsemblSeqProxy
    * <li>resolves an external identifier by looking up xref-ed gene ids</li>
    * <li>fetches the gene sequence</li>
    * <li>fetches features on the sequence</li>
-   * <li>identifies "transcript" features whose Parent is the requested gene</li>
+   * <li>identifies "transcript" features whose Parent is the requested
+   * gene</li>
    * <li>fetches the transcript sequence for each transcript</li>
    * <li>makes a mapping from the gene to each transcript</li>
    * <li>copies features from gene to transcript sequences</li>
@@ -191,7 +192,22 @@ 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 
@@ -298,8 +314,8 @@ public class EnsemblGene extends EnsemblSeqProxy
    *          the parent gene sequence, with features
    * @return
    */
-  SequenceI makeTranscript(SequenceFeature transcriptFeature,
-          AlignmentI al, SequenceI gene)
+  SequenceI makeTranscript(SequenceFeature transcriptFeature, AlignmentI al,
+          SequenceI gene)
   {
     String accId = getTranscriptId(transcriptFeature);
     if (accId == null)
@@ -348,7 +364,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       mappedFrom.add(new int[] { sf.getBegin(), sf.getEnd() });
     }
 
-    Sequence transcript = new Sequence(accId, seqChars, 1, transcriptLength);
+    Sequence transcript = new Sequence(accId, seqChars, 1,
+            transcriptLength);
 
     /*
      * Ensembl has gene name as transcript Name
@@ -411,6 +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
@@ -422,19 +445,18 @@ public class EnsemblGene extends EnsemblSeqProxy
     List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
 
     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.equals(parent))
       {
-        String parent = (String) sf.getValue(PARENT);
-        if (parentIdentifier.equals(parent))
-        {
-          transcriptFeatures.add(sf);
-        }
+        transcriptFeatures.add(sf);
       }
     }
 
@@ -551,8 +573,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       @Override
       public boolean isFeatureDisplayed(String type)
       {
-        return (so.isA(type, SequenceOntologyI.EXON) || so.isA(type,
-                SequenceOntologyI.SEQUENCE_VARIANT));
+        return (so.isA(type, SequenceOntologyI.EXON)
+                || so.isA(type, SequenceOntologyI.SEQUENCE_VARIANT));
       }
 
       @Override