X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblGene.java;h=919134ccb3e267b974c6135d7648666409f518eb;hb=17ff1f476e009b3a3c7e892e416edc2a4af8a2bc;hp=37c787b37222a61a29e186554ec346ecbe22bc88;hpb=7d9e8749e06f98fd14f72086faef8b6dd5c5adf4;p=jalview.git

diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java
index 37c787b..919134c 100644
--- 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;
@@ -97,6 +98,12 @@ 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
@@ -109,7 +116,8 @@ 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>
@@ -159,66 +167,43 @@ 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
    */
   List<String> getGeneIds(String accessions)
   {
-    List<String> geneIds = new ArrayList<String>();
+    List<String> geneIds = new ArrayList<>();
 
     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);
         }
       }
-      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 
-       */
       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);
           }
         }
       }
@@ -227,30 +212,6 @@ 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.
@@ -282,22 +243,20 @@ 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);
     }
   }
 
@@ -314,8 +273,8 @@ 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)
@@ -347,11 +306,12 @@ public class EnsemblGene extends EnsemblSeqProxy
     {
       splices = findFeatures(gene, SequenceOntologyI.CDS, parentId);
     }
+    SequenceFeatures.sortFeatures(splices, true);
 
     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)
     {
@@ -363,7 +323,8 @@ 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
@@ -392,11 +353,11 @@ 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());
-    cdna.transferFeatures(gene.getSequenceFeatures(),
+    cdna.transferFeatures(gene.getFeatures().getPositionalFeatures(),
             transcript.getDatasetSequence(), mapping, parentId);
 
     /*
@@ -426,6 +387,12 @@ 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
@@ -434,23 +401,21 @@ 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;
-    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);
       }
     }
 
@@ -485,7 +450,8 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
             SequenceOntologyI.GENE))
     {
-      String id = (String) sf.getValue(ID);
+      // NB features as gff use 'ID'; rest services return as 'id'
+      String id = (String) sf.getValue("ID");
       if ((GENE_PREFIX + accId).equals(id))
       {
         return true;
@@ -567,8 +533,8 @@ 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