Merge branch 'develop' into features/JAL-1793VCF

[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 0d1b554..4c43281 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblLookup.java
+++ b/src/jalview/ext/ensembl/EnsemblLookup.java
@@ -39,7 +39,9 @@ import org.json.simple.parser.JSONParser;
 import org.json.simple.parser.ParseException;
 
 /**
- * A client for the Ensembl lookup REST endpoint
+ * A client for the Ensembl /lookup REST endpoint, used to find the gene
+ * identifier given a gene, transcript or protein identifier, or to extract the
+ * species or chromosomal coordinates from the same service response
  * 
  * @author gmcarstairs
  */
@@ -47,14 +49,6 @@ public class EnsemblLookup extends EnsemblRestClient
 {
   private static final String SPECIES = "species";
 
-  private static final String PARENT = "Parent";
-
-  private static final String OBJECT_TYPE_TRANSLATION = "Translation";
-  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)
    */
@@ -89,17 +83,26 @@ 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);
@@ -128,20 +131,15 @@ public class EnsemblLookup extends EnsemblRestClient
   }
 
   /**
-   * Calls the Ensembl lookup REST endpoint and returns
-   * <ul>
-   * <li>the 'id' for the identifier if its type is "Gene"</li>
-   * <li>the 'Parent' if its type is 'Transcript'</li>
-   * <ul>
-   * If the type is 'Translation', does a recursive call to this method, passing
-   * in the 'Parent' (transcript id).
+   * 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 (String) getResult(identifier, br -> parseGeneId(br));
+    return (String) getResult(identifier, null, br -> parseGeneId(br));
   }
 
   /**
@@ -153,7 +151,8 @@ public class EnsemblLookup extends EnsemblRestClient
    */
   public String getSpecies(String identifier)
   {
-    return (String) getResult(identifier, br -> getAttribute(br, SPECIES));
+    return (String) getResult(identifier, null,
+            br -> getAttribute(br, SPECIES));
   }
 
   /**
@@ -161,10 +160,12 @@ public class EnsemblLookup extends EnsemblRestClient
    * response to the supplied parser
    * 
    * @param identifier
+   * @param objectType
+   *          (optional)
    * @param parser
    * @return
    */
-  protected Object getResult(String identifier,
+  protected Object getResult(String identifier, String objectType,
           Function<BufferedReader, Object> parser)
   {
     List<String> ids = Arrays.asList(new String[] { identifier });
@@ -172,7 +173,7 @@ public class EnsemblLookup extends EnsemblRestClient
     BufferedReader br = null;
     try
     {
-      URL url = getUrl(identifier);
+      URL url = getUrl(identifier, objectType);
       if (url != null)
       {
         br = getHttpResponse(url, ids);
@@ -224,7 +225,8 @@ public class EnsemblLookup extends EnsemblRestClient
    * Parses the JSON response and returns the gene identifier, or null if not
    * found. If the returned object_type is Gene, returns the id, if Transcript
    * returns the Parent. If it is Translation (peptide identifier), then the
-   * Parent is the transcript identifier, so we redo the search with this value.
+   * Parent is the transcript identifier, so we redo the search with this value,
+   * specifying that object_type should be Transcript.
    * 
    * @param br
    * @return
@@ -239,28 +241,20 @@ public class EnsemblLookup extends EnsemblRestClient
       String type = val.get(OBJECT_TYPE).toString();
       if (OBJECT_TYPE_GENE.equalsIgnoreCase(type))
       {
-        geneId = val.get(ID).toString();
+        // got the gene - just returns its id
+        geneId = val.get(JSON_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 = (String) getResult(transcriptId, OBJECT_TYPE_TRANSCRIPT,
+                reader -> parseGeneId(reader));
       }
     } catch (ParseException | IOException e)
     {
@@ -278,7 +272,8 @@ public class EnsemblLookup extends EnsemblRestClient
    */
   public GeneLociI getGeneLoci(String geneId)
   {
-    return (GeneLociI) getResult(geneId, br -> parseGeneLoci(br));
+    return (GeneLociI) getResult(geneId, OBJECT_TYPE_GENE,
+            br -> parseGeneLoci(br));
   }
 
   /**
@@ -341,6 +336,7 @@ public class EnsemblLookup extends EnsemblRestClient
             | NumberFormatException | ClassCastException e)
     {
       Cache.log.error("Error looking up gene loci: " + e.getMessage());
+      e.printStackTrace();
     }
     return null;
   }