From: Jim Procter <jprocter@issues.jalview.org>
Date: Sun, 2 Oct 2016 11:15:21 +0000 (+0100)
Subject: JAL-2210 discover candidate dataset sequences for an xref based on primaryDBRefs... 
X-Git-Tag: Release_2_10_0~20^2~16
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=639ac6c56cbbd596611ba7a2152380d6c9474090;p=jalview.git

JAL-2210 discover candidate dataset sequences for an xref based on primaryDBRefs first, and only fall back to sequence name match if none are found
---

diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java
index 6779b87..1ddeb20 100644
--- a/src/jalview/analysis/CrossRef.java
+++ b/src/jalview/analysis/CrossRef.java
@@ -637,9 +637,12 @@ public class CrossRef
   }
 
   /**
-   * Returns the first identical sequence in the dataset if any, else null
+   * Returns null or the first sequence in the dataset which is identical to
+   * xref.mapTo, and has a) a primary dbref matching xref, or if none found, the
+   * first one with an ID source|xrefacc
    * 
    * @param xref
+   *          with map and mapped-to sequence
    * @return
    */
   SequenceI findInDataset(DBRefEntry xref)
@@ -658,23 +661,37 @@ public class CrossRef
     {
       return dss;
     }
-    ;
+    DBRefEntry template = new DBRefEntry(xref.getSource(), null,
+            xref.getAccessionId());
+    /**
+     * remember the first ID match - in case we don't find a match to template
+     */
+    SequenceI firstIdMatch = null;
     for (SequenceI seq : dataset.getSequences())
     {
+      // first check primary refs.
+      List<DBRefEntry> match = DBRefUtils.searchRefs(seq.getPrimaryDBRefs()
+              .toArray(new DBRefEntry[0]), template);
+      if (match != null && match.size() == 1 && sameSequence(seq, dss))
+      {
+        return seq;
+      }
       /*
        * clumsy alternative to using SequenceIdMatcher which currently
        * returns sequences with a dbref to the matched accession id 
        * which we don't want
        */
-      if (name.equals(seq.getName()) || seq.getName().startsWith(name2))
+      if (firstIdMatch == null
+              && (name.equals(seq.getName()) || seq.getName().startsWith(
+                      name2)))
       {
         if (sameSequence(seq, dss))
         {
-          return seq;
+          firstIdMatch = seq;
         }
       }
     }
-    return null;
+    return firstIdMatch;
   }
 
   /**