JAL-2154 check that xref is a coding xref before adding sequence to retrieved xrefs

[jalview.git] / src / jalview / analysis / CrossRef.java

diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java
index 9932f2c..aefc6f8 100644 (file)
--- a/src/jalview/analysis/CrossRef.java
+++ b/src/jalview/analysis/CrossRef.java
@@ -222,6 +222,9 @@ public class CrossRef
       boolean found = false;
       DBRefEntry[] xrfs = DBRefUtils
               .selectDbRefs(!fromDna, dss.getDBRefs());
+      // ENST & ENSP comes in to both Protein and nucleotide, so we need to
+      // filter them
+      // out later.
       if ((xrfs == null || xrfs.length == 0) && dataset != null)
       {
         /*
@@ -249,11 +252,15 @@ public class CrossRef
       List<DBRefEntry> sourceRefs = DBRefUtils.searchRefsForSource(xrfs,
               source);
       Iterator<DBRefEntry> refIterator = sourceRefs.iterator();
+      // At this point, if we are retrieving Ensembl, we still don't filter out
+      // ENST when looking for protein crossrefs.
       while (refIterator.hasNext())
       {
         DBRefEntry xref = refIterator.next();
         found = false;
-        if (xref.hasMap())
+        // we're only interested in coding cross-references, not
+        // locus->transcript
+        if (xref.hasMap() && xref.getMap().getMap().isTripletMap())
         {
           SequenceI mappedTo = xref.getMap().getTo();
           if (mappedTo != null)
@@ -332,7 +339,9 @@ public class CrossRef
         {
           SequenceI matchedSeq = matcher.findIdMatch(xref.getSource() + "|"
                   + xref.getAccessionId());
-          if (matchedSeq != null)
+          // if there was a match, check it's at least the right type of
+          // molecule!
+          if (matchedSeq != null && matchedSeq.isProtein() == fromDna)
           {
             if (constructMapping(seq, matchedSeq, xref, cf, fromDna))
             {