JAL-4241 Fix non-standard residues replacement.

[jalview.git] / src / jalview / ws2 / actions / annotation / AnnotationJob.java

diff --git a/src/jalview/ws2/actions/annotation/AnnotationJob.java b/src/jalview/ws2/actions/annotation/AnnotationJob.java
index 467dafd..187c14c 100644 (file)
--- a/src/jalview/ws2/actions/annotation/AnnotationJob.java
+++ b/src/jalview/ws2/actions/annotation/AnnotationJob.java
@@ -71,18 +71,20 @@ public class AnnotationJob extends BaseJob
       String newName = SeqsetUtils.unique_name(seqences.size() + 1);
       if (namesMap != null)
         namesMap.put(newName, sq);
+      char[] seqChars = sq.getSequence(start, end + 1);
+      if (filterNonStandardResidues)
+        replaceNonStandardResidues(seqChars, Comparison.GAP_DASH, sq.isProtein());
       Sequence seq;
       if (submitGaps)
       {
-        seq = new Sequence(newName, sq.getSequenceAsString());
+        seq = new Sequence(newName, seqChars);
         updateResidueMap(residueMap, seq, filterNonStandardResidues);
       }
       else
       {
         // TODO: add ability to exclude hidden regions
         seq = new Sequence(newName,
-                AlignSeq.extractGaps(Comparison.GapChars,
-                        sq.getSequenceAsString(start, end + 1)));
+                AlignSeq.extractGaps(Comparison.GapChars, new String(seqChars)));
         // for annotation need to also record map to sequence start/end
         // position in range
         // then transfer back to original sequence on return.
@@ -112,6 +114,20 @@ public class AnnotationJob extends BaseJob
             minSize);
   }
 
+  private static void replaceNonStandardResidues(char[] seq, char replacement, boolean isProtein)
+  {
+    for (int i = 0; i < seq.length; i++)
+    {
+      char chr = seq[i];
+      if (isProtein
+          ? ResidueProperties.aaIndex[chr] >= 20
+          : ResidueProperties.nucleotideIndex[chr] >= 5)
+      {
+        seq[i] = replacement;
+      }
+    }
+  }
+
   private static void updateResidueMap(BitSet residueMap, SequenceI seq,
           boolean filterNonStandardResidues)
   {