Merge branch 'bug/JAL-2617mappedStopCodon' into develop

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

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index 2b9b9f9..90d9197 100644 (file)
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -2164,7 +2164,10 @@ public class AlignmentUtils
 
   /**
    * Returns a mapping from dna to protein by inspecting sequence features of
-   * type "CDS" on the dna.
+   * type "CDS" on the dna. A mapping is constructed if the total CDS feature
+   * length is 3 times the peptide length (optionally after dropping a trailing
+   * stop codon). This method does not check whether the CDS nucleotide sequence
+   * translates to the peptide sequence.
    * 
    * @param dnaSeq
    * @param proteinSeq
@@ -2176,6 +2179,15 @@ public class AlignmentUtils
     List<int[]> ranges = findCdsPositions(dnaSeq);
     int mappedDnaLength = MappingUtils.getLength(ranges);
 
+    /*
+     * if not a whole number of codons, something is wrong,
+     * abort mapping
+     */
+    if (mappedDnaLength % CODON_LENGTH > 0)
+    {
+      return null;
+    }
+
     int proteinLength = proteinSeq.getLength();
     int proteinStart = proteinSeq.getStart();
     int proteinEnd = proteinSeq.getEnd();
@@ -2199,8 +2211,12 @@ public class AlignmentUtils
     if (codesForResidues == (proteinLength + 1))
     {
       // assuming extra codon is for STOP and not in peptide
+      // todo: check trailing codon is indeed a STOP codon
       codesForResidues--;
+      mappedDnaLength -= CODON_LENGTH;
+      MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
     }
+
     if (codesForResidues == proteinLength)
     {
       proteinRange.add(new int[] { proteinStart, proteinEnd });
@@ -2211,7 +2227,7 @@ public class AlignmentUtils
 
   /**
    * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
-   * start/end positions of sequence features of type "CDS" (or a sub-type of
+   * [start, end] positions of sequence features of type "CDS" (or a sub-type of
    * CDS in the Sequence Ontology). The ranges are sorted into ascending start
    * position order, so this method is only valid for linear CDS in the same
    * sense as the protein product.
@@ -2230,7 +2246,6 @@ public class AlignmentUtils
       return result;
     }
     SequenceFeatures.sortFeatures(sfs, true);
-    int startPhase = 0;
 
     for (SequenceFeature sf : sfs)
     {
@@ -2248,7 +2263,7 @@ public class AlignmentUtils
        */
       int begin = sf.getBegin();
       int end = sf.getEnd();
-      if (result.isEmpty())
+      if (result.isEmpty() && phase > 0)
       {
         begin += phase;
         if (begin > end)
@@ -2263,16 +2278,6 @@ public class AlignmentUtils
     }
 
     /*
-     * remove 'startPhase' positions (usually 0) from the first range 
-     * so we begin at the start of a complete codon
-     */
-    if (!result.isEmpty())
-    {
-      // TODO JAL-2022 correctly model start phase > 0
-      result.get(0)[0] += startPhase;
-    }
-
-    /*
      * Finally sort ranges by start position. This avoids a dependency on 
      * keeping features in order on the sequence (if they are in order anyway,
      * the sort will have almost no work to do). The implicit assumption is CDS