Merge branch 'features/JAL-653_JAL-1766_htslib_refseqsupport' into develop

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

diff --git a/src/jalview/ext/ensembl/EnsemblProtein.java b/src/jalview/ext/ensembl/EnsemblProtein.java
index 8f23984..97796a5 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblProtein.java
+++ b/src/jalview/ext/ensembl/EnsemblProtein.java
@@ -8,20 +8,43 @@ import java.util.List;
 
 import com.stevesoft.pat.Regex;
 
+/**
+ * A client to fetch protein translated sequence for an Ensembl identifier
+ * 
+ * @author gmcarstairs
+ *
+ */
 public class EnsemblProtein extends EnsemblSeqProxy
 {
-  // TODO modify to accept other species e.g. ENSMUSPnnn
+  /*
+   * accepts ENSP 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(
-          "((ENSP|CCDS)[0-9.]{3,})");
+          "(ENS([A-Z]{3}|)P[0-9]{11}$)" + "|" + "(CCDS[0-9.]{3,}$)");
 
   private static final List<String> CROSSREFS = Arrays.asList(new String[] {
       "PDB", "Uniprot/SPTREMBL", "Uniprot/SWISSPROT" });
 
+  /**
+   * Default constructor (to use rest.ensembl.org)
+   */
   public EnsemblProtein()
   {
     super();
   }
 
+  /**
+   * Constructor given the target domain to fetch data from
+   * 
+   * @param d
+   */
+  public EnsemblProtein(String d)
+  {
+    super(d);
+  }
+
   @Override
   public String getDbName()
   {
@@ -87,4 +110,28 @@ public class EnsemblProtein extends EnsemblSeqProxy
     return ACCESSION_REGEX;
   }
 
+  /**
+   * Returns an accession id for a query, including conversion of ENST* to
+   * ENSP*. This supports querying for the protein sequence for a transcript
+   * (ENST identifier) and returning the ENSP identifier.
+   */
+  @Override
+  public String getAccessionIdFromQuery(String query)
+  {
+    String accId = super.getAccessionIdFromQuery(query);
+
+    /*
+     * ensure last character before (11) digits is P
+     * ENST00000288602 -> ENSP00000288602
+     * ENSMUST00000288602 -> ENSMUSP00000288602
+     */
+    if (accId != null && accId.length() >= 12)
+    {
+      char[] chars = accId.toCharArray();
+      chars[chars.length - 12] = 'P';
+      accId = new String(chars);
+    }
+    return accId;
+  }
+
 }