JAL-2232 JAL-2154 update dbrefs to newly created dataset sequence

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

diff --git a/src/jalview/ext/ensembl/EnsemblSequenceFetcher.java b/src/jalview/ext/ensembl/EnsemblSequenceFetcher.java
index f1b96e2..dd1739b 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblSequenceFetcher.java
+++ b/src/jalview/ext/ensembl/EnsemblSequenceFetcher.java
@@ -9,13 +9,24 @@ import com.stevesoft.pat.Regex;
  * A base class for Ensembl sequence fetchers
  * 
  * @author gmcarstairs
- *
  */
-public abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
+abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
 {
   /*
-   * possible values for the 'feature' parameter of the REST overlap endpoint
-   * @see 
+   * accepts ENSG/T/E/P with 11 digits
+   * or ENSMUSP or similar for other species
+   * or CCDSnnnnn.nn with at least 3 digits
+   */
+  private static final Regex ACCESSION_REGEX = new Regex(
+          "(ENS([A-Z]{3}|)[GTEP]{1}[0-9]{11}$)" + "|" + "(CCDS[0-9.]{3,}$)");
+
+  protected static final String ENSEMBL_GENOMES_REST = "http://rest.ensemblgenomes.org";
+
+  protected static final String ENSEMBL_REST = "http://rest.ensembl.org";
+
+  /*
+   * possible values for the 'feature' parameter of the /overlap REST service
+   * @see http://rest.ensembl.org/documentation/info/overlap_id
    */
   protected enum EnsemblFeatureType
   {
@@ -24,17 +35,17 @@ public abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
     constrained, regulatory
   }
 
+  private String domain = ENSEMBL_REST;
+
   @Override
   public String getDbSource()
   {
     // NB ensure Uniprot xrefs are canonicalised from "Ensembl" to "ENSEMBL"
-    return DBRefSource.ENSEMBL; // "ENSEMBL"
-  }
-
-  @Override
-  public String getDbVersion()
-  {
-    return "0";
+    if (ENSEMBL_GENOMES_REST.equals(getDomain()))
+    {
+      return DBRefSource.ENSEMBLGENOMES;
+    }
+    return DBRefSource.ENSEMBL;
   }
 
   @Override
@@ -43,10 +54,16 @@ public abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
     return " ";
   }
 
+  /**
+   * Ensembl accession are ENST + 11 digits for human transcript, ENSG for human
+   * gene. Other species insert 3 letters e.g. ENSMUST..., ENSMUSG...
+   * 
+   * @see http://www.ensembl.org/Help/View?id=151
+   */
   @Override
   public Regex getAccessionValidator()
   {
-    return new Regex("((ENSP|ENST|ENSG|CCDS)[0-9.]{3,})");
+    return ACCESSION_REGEX;
   }
 
   @Override
@@ -77,4 +94,20 @@ public abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
   {
     return true;
   }
+
+  /**
+   * Returns the domain name to query e.g. http://rest.ensembl.org or
+   * http://rest.ensemblgenomes.org
+   * 
+   * @return
+   */
+  protected String getDomain()
+  {
+    return domain;
+  }
+
+  protected void setDomain(String d)
+  {
+    domain = d;
+  }
 }