From: Mungo Carstairs <gmungoc@gmail.com>
Date: Tue, 18 Feb 2020 12:13:13 +0000 (+0000)
Subject: JAL-3536 JAL-3187 don't create duplicate CDS-to-protein mappings
X-Git-Tag: Develop-2_11_2_0-d20201215~98^2
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=refs%2Fheads%2Fbug%2FJAL-3536getCrossrefTwice;p=jalview.git

JAL-3536 JAL-3187 don't create duplicate CDS-to-protein mappings
---

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index fdf66d0..2b88cb0 100644
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -20,8 +20,6 @@
  */
 package jalview.analysis;
 
-import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
-
 import jalview.commands.RemoveGapColCommand;
 import jalview.datamodel.AlignedCodon;
 import jalview.datamodel.AlignedCodonFrame;
@@ -38,7 +36,6 @@ import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceGroup;
 import jalview.datamodel.SequenceI;
 import jalview.datamodel.features.SequenceFeatures;
-import jalview.io.gff.Gff3Helper;
 import jalview.io.gff.SequenceOntologyI;
 import jalview.schemes.ResidueProperties;
 import jalview.util.Comparison;
@@ -46,10 +43,7 @@ import jalview.util.DBRefUtils;
 import jalview.util.IntRangeComparator;
 import jalview.util.MapList;
 import jalview.util.MappingUtils;
-import jalview.util.StringUtils;
 
-import java.io.UnsupportedEncodingException;
-import java.net.URLEncoder;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collection;
@@ -1737,15 +1731,8 @@ public class AlignmentUtils
 
           cdsSeqs.add(cdsSeq);
 
-          if (!dataset.getSequences().contains(cdsSeqDss))
-          {
-            // check if this sequence is a newly created one
-            // so needs adding to the dataset
-            dataset.addSequence(cdsSeqDss);
-          }
-
           /*
-           * add a mapping from CDS to the (unchanged) mapped to range
+           * build the mapping from CDS to protein
            */
           List<int[]> cdsRange = Collections
                   .singletonList(new int[]
@@ -1754,16 +1741,26 @@ public class AlignmentUtils
           MapList cdsToProteinMap = new MapList(cdsRange,
                   mapList.getToRanges(), mapList.getFromRatio(),
                   mapList.getToRatio());
-          AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
-          cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
-                  cdsToProteinMap);
 
-          /*
-           * guard against duplicating the mapping if repeating this action
-           */
-          if (!mappings.contains(cdsToProteinMapping))
+          if (!dataset.getSequences().contains(cdsSeqDss))
           {
-            mappings.add(cdsToProteinMapping);
+            /*
+             * if this sequence is a newly created one, add it to the dataset
+             * and made a CDS to protein mapping (if sequence already exists,
+             * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
+             */
+            dataset.addSequence(cdsSeqDss);
+            AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
+            cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
+                  cdsToProteinMap);
+
+            /*
+             * guard against duplicating the mapping if repeating this action
+             */
+            if (!mappings.contains(cdsToProteinMapping))
+            {
+              mappings.add(cdsToProteinMapping);
+            }
           }
 
           propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
@@ -2576,6 +2573,13 @@ public class AlignmentUtils
     {
       List<SequenceI> alignedSequences = alignedDatasets
               .get(seq.getDatasetSequence());
+      if (alignedSequences.isEmpty())
+      {
+        /*
+         * defensive check - shouldn't happen! (JAL-3536)
+         */
+        continue;
+      }
       SequenceI alignedSeq = alignedSequences.get(0);
 
       /*