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

[jalview.git] / src / jalview / ext / ensembl / EnsemblLookup.java

diff --git a/src/jalview/ext/ensembl/EnsemblLookup.java b/src/jalview/ext/ensembl/EnsemblLookup.java
index 31da9c0..92763a1 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblLookup.java
+++ b/src/jalview/ext/ensembl/EnsemblLookup.java
@@ -34,22 +34,13 @@ import org.json.simple.parser.JSONParser;
 import org.json.simple.parser.ParseException;
 
 /**
- * A client for the Ensembl lookup REST endpoint; used to find the Parent gene
- * identifier given a transcript identifier.
+ * A client for the Ensembl lookup REST endpoint, used to find the gene
+ * identifier given a gene, transcript or protein identifier.
  * 
  * @author gmcarstairs
- *
  */
 public class EnsemblLookup extends EnsemblRestClient
 {
-
-  private static final String OBJECT_TYPE_TRANSLATION = "Translation";
-  private static final String PARENT = "Parent";
-  private static final String OBJECT_TYPE_TRANSCRIPT = "Transcript";
-  private static final String ID = "id";
-  private static final String OBJECT_TYPE_GENE = "Gene";
-  private static final String OBJECT_TYPE = "object_type";
-
   /**
    * Default constructor (to use rest.ensembl.org)
    */
@@ -84,17 +75,26 @@ public class EnsemblLookup extends EnsemblRestClient
   protected URL getUrl(List<String> ids) throws MalformedURLException
   {
     String identifier = ids.get(0);
-    return getUrl(identifier);
+    return getUrl(identifier, null);
   }
 
   /**
+   * Gets the url for lookup of the given identifier, optionally with objectType
+   * also specified in the request
+   * 
    * @param identifier
+   * @param objectType
    * @return
    */
-  protected URL getUrl(String identifier)
+  protected URL getUrl(String identifier, String objectType)
   {
     String url = getDomain() + "/lookup/id/" + identifier
             + CONTENT_TYPE_JSON;
+    if (objectType != null)
+    {
+      url += "&" + OBJECT_TYPE + "=" + objectType;
+    }
+
     try
     {
       return new URL(url);
@@ -123,20 +123,34 @@ public class EnsemblLookup extends EnsemblRestClient
   }
 
   /**
+   * Returns the gene id related to the given identifier, which may be for a
+   * gene, transcript or protein
+   * 
+   * @param identifier
+   * @return
+   */
+  public String getGeneId(String identifier)
+  {
+    return getGeneId(identifier, null);
+  }
+
+  /**
    * Calls the Ensembl lookup REST endpoint and retrieves the 'Parent' for the
    * given identifier, or null if not found
    * 
    * @param identifier
+   * @param objectType
+   *          (optional)
    * @return
    */
-  public String getGeneId(String identifier)
+  public String getGeneId(String identifier, String objectType)
   {
     List<String> ids = Arrays.asList(new String[] { identifier });
 
     BufferedReader br = null;
     try
     {
-      URL url = getUrl(identifier);
+      URL url = getUrl(identifier, objectType);
       if (url != null)
       {
         br = getHttpResponse(url, ids);
@@ -181,28 +195,19 @@ public class EnsemblLookup extends EnsemblRestClient
       String type = val.get(OBJECT_TYPE).toString();
       if (OBJECT_TYPE_GENE.equalsIgnoreCase(type))
       {
+        // got the gene - just returns its id
         geneId = val.get(ID).toString();
       }
       else if (OBJECT_TYPE_TRANSCRIPT.equalsIgnoreCase(type))
       {
+        // got the transcript - return its (Gene) Parent
         geneId = val.get(PARENT).toString();
       }
       else if (OBJECT_TYPE_TRANSLATION.equalsIgnoreCase(type))
       {
+        // got the protein - get its Parent, restricted to type Transcript
         String transcriptId = val.get(PARENT).toString();
-        try
-        {
-          geneId = getGeneId(transcriptId);
-        } catch (StackOverflowError e)
-        {
-          /*
-           * unlikely data condition error!
-           */
-          System.err
-                  .println("** Ensembl lookup "
-                          + getUrl(transcriptId).toString()
-                          + " looping on Parent!");
-        }
+        geneId = getGeneId(transcriptId, OBJECT_TYPE_TRANSCRIPT);
       }
     } catch (ParseException e)
     {