From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Tue, 21 Jun 2016 09:08:48 +0000 (+0100)
Subject: JAL-2113 tidy up canonicalisation of Uniprot xrefs in EMBL data
X-Git-Tag: Release_2_10_0~140^2~5^2~49^2~1
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=f4debc094f95109d1eb0518914d8ada444cd3f5d;p=jalview.git

JAL-2113 tidy up canonicalisation of Uniprot xrefs in EMBL data
---

diff --git a/src/jalview/datamodel/xdb/embl/EmblEntry.java b/src/jalview/datamodel/xdb/embl/EmblEntry.java
index ac0d339..43581da 100644
--- a/src/jalview/datamodel/xdb/embl/EmblEntry.java
+++ b/src/jalview/datamodel/xdb/embl/EmblEntry.java
@@ -210,13 +210,14 @@ public class EmblEntry
       }
     }
 
+    SequenceIdMatcher matcher = new SequenceIdMatcher(peptides);
     try
     {
       for (EmblFeature feature : features)
       {
         if (FeatureProperties.isCodingFeature(sourceDb, feature.getName()))
         {
-          parseCodingFeature(feature, sourceDb, dna, peptides);
+          parseCodingFeature(feature, sourceDb, dna, peptides, matcher);
         }
       }
     } catch (Exception e)
@@ -244,9 +245,11 @@ 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);
 
@@ -256,7 +259,6 @@ 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
diff --git a/test/jalview/datamodel/xdb/embl/EmblEntryTest.java b/test/jalview/datamodel/xdb/embl/EmblEntryTest.java
index e8760bd..3de5e3f 100644
--- a/test/jalview/datamodel/xdb/embl/EmblEntryTest.java
+++ b/test/jalview/datamodel/xdb/embl/EmblEntryTest.java
@@ -3,6 +3,7 @@ package jalview.datamodel.xdb.embl;
 import static org.testng.AssertJUnit.assertEquals;
 import static org.testng.AssertJUnit.assertSame;
 
+import jalview.analysis.SequenceIdMatcher;
 import jalview.datamodel.DBRefEntry;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceI;
@@ -37,6 +38,7 @@ public class EmblEntryTest
     // not the whole sequence but enough for this test...
     SequenceI dna = new Sequence("J03321", "GGATCCGTAAGTTAGACGAAATT");
     List<SequenceI> peptides = new ArrayList<SequenceI>();
+    SequenceIdMatcher matcher = new SequenceIdMatcher(peptides);
     EmblFile ef = EmblTestHelper.getEmblFile();
 
     /*
@@ -48,7 +50,7 @@ public class EmblEntryTest
     {
       if ("CDS".equals(feature.getName()))
       {
-        testee.parseCodingFeature(feature, "EMBL", dna, peptides);
+        testee.parseCodingFeature(feature, "EMBL", dna, peptides, matcher);
       }
     }