JAL-2051 better checking of retrieved / duplicated accession ids

[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 84b5dcf..4dd1bba 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -11,7 +11,6 @@ import jalview.io.gff.SequenceOntologyI;
 import jalview.schemes.FeatureColourAdapter;
 import jalview.schemes.FeatureSettingsAdapter;
 import jalview.util.MapList;
-import jalview.util.StringUtils;
 
 import java.awt.Color;
 import java.io.UnsupportedEncodingException;
@@ -108,47 +107,84 @@ public class EnsemblGene extends EnsemblSeqProxy
   public AlignmentI getSequenceRecords(String query) throws Exception
   {
     /*
-     * if given a transcript id, look up its gene parent
+     * convert to a non-duplicated list of gene identifiers
      */
-    if (isTranscriptIdentifier(query))
+    List<String> geneIds = getGeneIds(query);
+
+    AlignmentI al = null;
+    for (String geneId : geneIds)
     {
-      query = new EnsemblLookup(getDomain()).getParent(query);
-      if (query == null)
+      /*
+       * fetch the gene sequence(s) with features and xrefs
+       */
+      AlignmentI geneAlignment = super.getSequenceRecords(geneId);
+
+      if (geneAlignment.getHeight() == 1)
       {
-        return null;
+        getTranscripts(geneAlignment, geneId);
+      }
+      if (al == null)
+      {
+        al = geneAlignment;
+      }
+      else
+      {
+        al.append(geneAlignment);
       }
     }
+    return al;
+  }
 
-    /*
-     * if given a gene or other external name, lookup and fetch 
-     * the corresponding gene for all model organisms 
-     */
-    if (!isGeneIdentifier(query))
+  /**
+   * Converts a query, which may contain one or more gene or transcript
+   * identifiers, into a non-redundant list of gene identifiers.
+   * 
+   * @param accessions
+   * @return
+   */
+  List<String> getGeneIds(String accessions)
+  {
+    List<String> geneIds = new ArrayList<String>();
+
+    for (String acc : accessions.split(getAccessionSeparator()))
     {
-      List<String> geneIds = new EnsemblSymbol(getDomain()).getIds(query);
-      if (geneIds.isEmpty())
+      if (isGeneIdentifier(acc))
       {
-        return null;
+        if (!geneIds.contains(acc))
+        {
+          geneIds.add(acc);
+        }
       }
-      String theIds = StringUtils.listToDelimitedString(geneIds,
-              getAccessionSeparator());
-      return getSequenceRecords(theIds);
-    }
 
-    /*
-     * fetch the gene sequence(s) with features and xrefs
-     */
-    AlignmentI al = super.getSequenceRecords(query);
+      /*
+       * if given a transcript id, look up its gene parent
+       */
+      else if (isTranscriptIdentifier(acc))
+      {
+        String geneId = new EnsemblLookup(getDomain()).getParent(acc);
+        if (geneId != null && !geneIds.contains(geneId))
+        {
+          geneIds.add(geneId);
+        }
+      }
 
-    /*
-     * if we retrieved a single gene, get its transcripts as well
-     */
-    if (al.getHeight() == 1)
-    {
-      getTranscripts(al, query);
+      /*
+       * if given a gene or other external name, lookup and fetch 
+       * the corresponding gene for all model organisms 
+       */
+      else
+      {
+        List<String> ids = new EnsemblSymbol(getDomain()).getIds(acc);
+        for (String geneId : ids)
+        {
+          if (!geneIds.contains(geneId))
+          {
+            geneIds.add(geneId);
+          }
+        }
+      }
     }
-
-    return al;
+    return geneIds;
   }
 
   /**
@@ -551,10 +587,4 @@ public class EnsemblGene extends EnsemblSeqProxy
     };
   }
 
-  @Override
-  public int getMaximumQueryCount()
-  {
-    return 1;
-  }
-
 }