JAL-2679 use object_type=Transcript for lookup of Parent for Protein

[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 ad01324..919134c 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -98,6 +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
@@ -161,66 +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);
           }
         }
       }
@@ -229,30 +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.
@@ -352,7 +311,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     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)
     {
@@ -394,7 +353,7 @@ 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());
@@ -442,7 +401,7 @@ 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;
 
@@ -491,7 +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;