X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FAlignmentUtils.java;h=c6cc5af439546860e7e1089acf1556a83fa620d3;hb=refs%2Fheads%2Ffeature%2FJAL-2738GeneLocus;hp=3d22115dc346eef6635e6ccf162b447e1921a439;hpb=a48b78b602cf3df82ce2821dc3520f7248f3cdf9;p=jalview.git

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index 3d22115..c6cc5af 100644
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -465,7 +465,7 @@ public class AlignmentUtils
     {
       String lastCodon = String.valueOf(cdnaSeqChars,
               cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
-      for (String stop : ResidueProperties.STOP)
+      for (String stop : ResidueProperties.STOP_CODONS)
       {
         if (lastCodon.equals(stop))
         {
@@ -536,7 +536,8 @@ public class AlignmentUtils
        * allow * in protein to match untranslatable in dna
        */
       final char aaRes = aaSeqChars[aaPos];
-      if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
+      if ((translated == null || ResidueProperties.STOP.equals(translated))
+              && aaRes == '*')
       {
         continue;
       }
@@ -568,7 +569,8 @@ public class AlignmentUtils
     if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
     {
       String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
-      if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
+      if (ResidueProperties.STOP
+              .equals(ResidueProperties.codonTranslate(codon)))
       {
         return true;
       }
@@ -1865,7 +1867,7 @@ public class AlignmentUtils
       return;
     }
 
-    MapList newMap = targetToFrom.traverse(fromLoci.getMap());
+    MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
 
     if (newMap != null)
     {
@@ -1886,7 +1888,7 @@ public class AlignmentUtils
    * @param seqMappings
    *          the set of mappings involving dnaSeq
    * @param aMapping
-   *          an initial candidate from seqMappings
+   *          a transcript-to-peptide mapping
    * @return
    */
   static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
@@ -1911,7 +1913,15 @@ public class AlignmentUtils
     if (mappedFromLength == dnaLength
             || mappedFromLength == dnaLength - CODON_LENGTH)
     {
-      return seqDss;
+      /*
+       * if sequence has CDS features, this is a transcript with no UTR
+       * - do not take this as the CDS sequence! (JAL-2789)
+       */
+      if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
+              .isEmpty())
+      {
+        return seqDss;
+      }
     }
 
     /*
@@ -1936,10 +1946,12 @@ public class AlignmentUtils
           {
             /*
             * found a 3:1 mapping to the protein product which covers
-            * the whole dna sequence i.e. is from CDS; finally check it
-            * is from the dna start sequence
+            * the whole dna sequence i.e. is from CDS; finally check the CDS
+            * is mapped from the given dna start sequence
             */
             SequenceI cdsSeq = map.getFromSeq();
+            // todo this test is weak if seqMappings contains multiple mappings;
+            // we get away with it if transcript:cds relationship is 1:1
             List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
                     .findMappingsForSequence(cdsSeq, seqMappings);
             if (!dnaToCdsMaps.isEmpty())
@@ -2054,9 +2066,11 @@ public class AlignmentUtils
   protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
           SequenceI contig, SequenceI proteinProduct, Mapping mapping)
   {
+
     // gather direct refs from contig congruent with mapping
     List<DBRefEntry> direct = new ArrayList<>();
     HashSet<String> directSources = new HashSet<>();
+
     if (contig.getDBRefs() != null)
     {
       for (DBRefEntry dbr : contig.getDBRefs())
@@ -2233,12 +2247,13 @@ public class AlignmentUtils
     int mappedDnaLength = MappingUtils.getLength(ranges);
 
     /*
-     * if not a whole number of codons, something is wrong,
-     * abort mapping
+     * if not a whole number of codons, truncate mapping
      */
-    if (mappedDnaLength % CODON_LENGTH > 0)
+    int codonRemainder = mappedDnaLength % CODON_LENGTH;
+    if (codonRemainder > 0)
     {
-      return null;
+      mappedDnaLength -= codonRemainder;
+      MappingUtils.removeEndPositions(codonRemainder, ranges);
     }
 
     int proteinLength = proteinSeq.getLength();
@@ -2428,11 +2443,14 @@ public class AlignmentUtils
         {
           for (String base : alleles.split(","))
           {
-            if (!base1.equals(base))
+            if (!base1.equalsIgnoreCase(base))
             {
-              String codon = base + base2 + base3;
+              String codon = base.toUpperCase() + base2.toLowerCase()
+                      + base3.toLowerCase();
+              String canonical = base1.toUpperCase() + base2.toLowerCase()
+                      + base3.toLowerCase();
               if (addPeptideVariant(peptide, peptidePos, residue, var,
-                      codon))
+                      codon, canonical))
               {
                 count++;
               }
@@ -2454,11 +2472,14 @@ public class AlignmentUtils
         {
           for (String base : alleles.split(","))
           {
-            if (!base2.equals(base))
+            if (!base2.equalsIgnoreCase(base))
             {
-              String codon = base1 + base + base3;
+              String codon = base1.toLowerCase() + base.toUpperCase()
+                      + base3.toLowerCase();
+              String canonical = base1.toLowerCase() + base2.toUpperCase()
+                      + base3.toLowerCase();
               if (addPeptideVariant(peptide, peptidePos, residue, var,
-                      codon))
+                      codon, canonical))
               {
                 count++;
               }
@@ -2480,11 +2501,14 @@ public class AlignmentUtils
         {
           for (String base : alleles.split(","))
           {
-            if (!base3.equals(base))
+            if (!base3.equalsIgnoreCase(base))
             {
-              String codon = base1 + base2 + base;
+              String codon = base1.toLowerCase() + base2.toLowerCase()
+                      + base.toUpperCase();
+              String canonical = base1.toLowerCase() + base2.toLowerCase()
+                      + base3.toUpperCase();
               if (addPeptideVariant(peptide, peptidePos, residue, var,
-                      codon))
+                      codon, canonical))
               {
                 count++;
               }
@@ -2498,20 +2522,22 @@ public class AlignmentUtils
   }
 
   /**
-   * Helper method that adds a peptide variant feature, provided the given codon
-   * translates to a value different to the current residue (is a non-synonymous
-   * variant). ID and clinical_significance attributes of the dna variant (if
-   * present) are copied to the new feature.
+   * Helper method that adds a peptide variant feature. ID and
+   * clinical_significance attributes of the dna variant (if present) are copied
+   * to the new feature.
    * 
    * @param peptide
    * @param peptidePos
    * @param residue
    * @param var
    * @param codon
+   *          the variant codon e.g. aCg
+   * @param canonical
+   *          the 'normal' codon e.g. aTg
    * @return true if a feature was added, else false
    */
   static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
-          String residue, DnaVariant var, String codon)
+          String residue, DnaVariant var, String codon, String canonical)
   {
     /*
      * get peptide translation of codon e.g. GAT -> D
@@ -2526,12 +2552,16 @@ public class AlignmentUtils
     {
       return false;
     }
-    String desc = codon;
+    String desc = canonical + "/" + codon;
     String featureType = "";
     if (trans.equals(residue))
     {
       featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
     }
+    else if (ResidueProperties.STOP.equals(trans))
+    {
+      featureType = SequenceOntologyI.STOP_GAINED;
+    }
     else
     {
       String residue3Char = StringUtils