X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2Fxdb%2Fembl%2FEmblEntry.java;h=5409d5b177066c80db96805504fd76308f129c5e;hb=0ac97c219bf88278f77306a5695e8bd9d9ca9179;hp=b7509735547418c703c2ac950cfbd48c18d167a3;hpb=627bafea52e4d5702bed00725afca1e0d17f8f60;p=jalview.git

diff --git a/src/jalview/datamodel/xdb/embl/EmblEntry.java b/src/jalview/datamodel/xdb/embl/EmblEntry.java
index b750973..5409d5b 100644
--- a/src/jalview/datamodel/xdb/embl/EmblEntry.java
+++ b/src/jalview/datamodel/xdb/embl/EmblEntry.java
@@ -197,7 +197,11 @@ public class EmblEntry
     // dbref
     retrievedref.setMap(new Mapping(null, new int[] { 1, dna.getLength() },
             new int[] { 1, dna.getLength() }, 1, 1));
-    // TODO: transform EMBL Database refs to canonical form
+
+
+    /*
+     * transform EMBL Database refs to canonical form
+     */
     if (dbRefs != null)
     {
       for (DBRefEntry dbref : dbRefs)
@@ -207,24 +211,14 @@ public class EmblEntry
       }
     }
 
+    SequenceIdMatcher matcher = new SequenceIdMatcher(peptides);
     try
     {
       for (EmblFeature feature : features)
       {
-        if (feature.dbRefs != null)
-        {
-          for (DBRefEntry dbref : feature.dbRefs)
-          {
-            /*
-             * convert UniProtKB/Swiss-Prot to UNIPROT
-             */
-            dbref.setSource(DBRefUtils.getCanonicalName(dbref.getSource()));
-            dna.addDBRef(dbref);
-          }
-        }
         if (FeatureProperties.isCodingFeature(sourceDb, feature.getName()))
         {
-          parseCodingFeature(feature, sourceDb, dna, peptides);
+          parseCodingFeature(feature, sourceDb, dna, peptides, matcher);
         }
       }
     } catch (Exception e)
@@ -252,9 +246,11 @@ public class EmblEntry
    *          parent dna sequence for this record
    * @param peptides
    *          list of protein product sequences for Embl entry
+   * @param matcher
+   *          helper to match xrefs in already retrieved sequences
    */
   void parseCodingFeature(EmblFeature feature, String sourceDb,
-          SequenceI dna, List<SequenceI> peptides)
+          SequenceI dna, List<SequenceI> peptides, SequenceIdMatcher matcher)
   {
     boolean isEmblCdna = sourceDb.equals(DBRefSource.EMBLCDS);
 
@@ -264,7 +260,6 @@ public class EmblEntry
     String prname = "";
     String prid = null;
     Map<String, String> vals = new Hashtable<String, String>();
-    SequenceIdMatcher matcher = new SequenceIdMatcher(peptides);
 
     /*
      * codon_start 1/2/3 in EMBL corresponds to phase 0/1/2 in CDS
@@ -288,13 +283,13 @@ public class EmblEntry
         }
         else if (qname.equals("protein_id"))
         {
-          prid = q.getValues()[0];
+          prid = q.getValues()[0].trim();
         }
         else if (qname.equals("codon_start"))
         {
           try
           {
-            codonStart = Integer.parseInt(q.getValues()[0]);
+            codonStart = Integer.parseInt(q.getValues()[0].trim());
           } catch (NumberFormatException e)
           {
             System.err.println("Invalid codon_start in XML for "
@@ -304,7 +299,7 @@ public class EmblEntry
         else if (qname.equals("product"))
         {
           // sometimes name is returned e.g. for V00488
-          prname = q.getValues()[0];
+          prname = q.getValues()[0].trim();
         }
         else
         {
@@ -426,13 +421,17 @@ public class EmblEntry
     }
 
     /*
-     * add mappings for Uniprot xrefs
+     * add dbRefs to sequence, and mappings for Uniprot xrefs
      */
     if (feature.dbRefs != null)
     {
       boolean mappingUsed = false;
       for (DBRefEntry ref : feature.dbRefs)
       {
+        /*
+         * ensure UniProtKB/Swiss-Prot converted to UNIPROT
+         */
+        ref.setSource(DBRefUtils.getCanonicalName(ref.getSource()));
         if (ref.getSource().equals(DBRefSource.UNIPROT))
         {
           String proteinSeqName = DBRefSource.UNIPROT + "|"
@@ -488,6 +487,7 @@ public class EmblEntry
             }
           }
         }
+        dna.addDBRef(ref);
       }
       if (noProteinDbref && product != null)
       {