Merge remote-tracking branch 'origin/develop' into imp/JAL-2774

[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..50dfa90 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>
@@ -160,8 +161,9 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Converts a query, which may contain one or more gene or transcript
-   * identifiers, into a non-redundant list of gene identifiers.
+   * Converts a query, which may contain one or more gene, transcript, or
+   * external (to Ensembl) identifiers, into a non-redundant list of gene
+   * identifiers.
    * 
    * @param accessions
    * @return
@@ -172,39 +174,30 @@ public class EnsemblGene extends EnsemblSeqProxy
 
     for (String acc : accessions.split(getAccessionSeparator()))
     {
-      if (isGeneIdentifier(acc))
-      {
-        if (!geneIds.contains(acc))
-        {
-          geneIds.add(acc);
-        }
-      }
-
       /*
-       * if given a transcript id, look up its gene parent
+       * First try lookup as an Ensembl (gene or transcript) identifier
        */
-      else if (isTranscriptIdentifier(acc))
+      String geneId = new EnsemblLookup(getDomain()).getGeneId(acc);
+      if (geneId != null)
       {
-        String geneId = new EnsemblLookup(getDomain()).getParent(acc);
-        if (geneId != null && !geneIds.contains(geneId))
+        if (!geneIds.contains(geneId))
         {
           geneIds.add(geneId);
         }
       }
-
-      /*
-       * if given a gene or other external name, lookup and fetch 
-       * the corresponding gene for all model organisms 
-       */
       else
       {
+        /*
+         * if given a gene or other external name, lookup and fetch 
+         * the corresponding gene for all model organisms 
+         */
         List<String> ids = new EnsemblSymbol(getDomain(), getDbSource(),
-                getDbVersion()).getIds(acc);
-        for (String geneId : ids)
+                getDbVersion()).getGeneIds(acc);
+        for (String id : ids)
         {
-          if (!geneIds.contains(geneId))
+          if (!geneIds.contains(id))
           {
-            geneIds.add(geneId);
+            geneIds.add(id);
           }
         }
       }
@@ -213,30 +206,6 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Attempts to get Ensembl stable identifiers for model organisms for a gene
-   * name by calling the xrefs symbol REST service to resolve the gene name.
-   * 
-   * @param query
-   * @return
-   */
-  protected String getGeneIdentifiersForName(String query)
-  {
-    List<String> ids = new EnsemblSymbol(getDomain(), getDbSource(),
-            getDbVersion()).getIds(query);
-    if (ids != null)
-    {
-      for (String id : ids)
-      {
-        if (isGeneIdentifier(id))
-        {
-          return id;
-        }
-      }
-    }
-    return null;
-  }
-
-  /**
    * Constructs all transcripts for the gene, as identified by "transcript"
    * features whose Parent is the requested gene. The coding transcript
    * sequences (i.e. with introns omitted) are added to the alignment.
@@ -298,8 +267,8 @@ 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 +317,8 @@ 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 +381,12 @@ 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 +398,18 @@ 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 +526,8 @@ 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