JAL-2154 don’t synthesise multiple CDS|<Acc> sequences when one is already available...

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

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index c4e51b1..b0a2269 100644 (file)
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -22,7 +22,6 @@ package jalview.analysis;
 
 import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
 
-import jalview.api.DBRefEntryI;
 import jalview.datamodel.AlignedCodon;
 import jalview.datamodel.AlignedCodonFrame;
 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
@@ -1681,11 +1680,20 @@ public class AlignmentUtils
            * didn't find mapped CDS sequence - construct it and add
            * its dataset sequence to the dataset
            */
-          cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping);
-          SequenceI cdsSeqDss = cdsSeq.createDatasetSequence();
+          cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
+                  dataset).deriveSequence();
+          // cdsSeq has a name constructed as CDS|<dbref>
+          // <dbref> will be either the accession for the coding sequence,
+          // marked in the /via/ dbref to the protein product accession
+          // or it will be the original nucleotide accession.
+          SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
+
           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);
           }
 
@@ -1731,12 +1739,37 @@ public class AlignmentUtils
            * same source and accession, so need a different accession for
            * the CDS from the dna sequence
            */
-          DBRefEntryI dnaRef = dnaDss.getSourceDBRef();
-          if (dnaRef != null)
+          
+          // specific use case:
+          // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
+          // ENSG02, ENSG03, with transcripts and products similarly named.
+          // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
+          
+          // JBPNote: ?? can't actually create an example that demonstrates we
+          // need to
+          // synthesize an xref.
+          
+          for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
           {
+            // creates a complementary cross-reference to the source sequence's
+            // primary reference.
+
+            DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
+                    primRef.getSource() + ":" + primRef.getVersion(),
+                    primRef.getAccessionId());
+            cdsCrossRef
+                    .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
+            cdsSeqDss.addDBRef(cdsCrossRef);
+
+            // problem here is that the cross-reference is synthesized -
+            // cdsSeq.getName() may be like 'CDS|dnaaccession' or
+            // 'CDS|emblcdsacc'
             // assuming cds version same as dna ?!?
-            DBRefEntry proteinToCdsRef = new DBRefEntry(dnaRef.getSource(),
-                    dnaRef.getVersion(), cdsSeq.getName());
+
+            DBRefEntry proteinToCdsRef = new DBRefEntry(
+                    primRef.getSource(), primRef.getVersion(),
+                    cdsSeq.getName());
+            //
             proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
                     .getInverse()));
             proteinProduct.addDBRef(proteinToCdsRef);
@@ -1842,9 +1875,14 @@ public class AlignmentUtils
    * 
    * @param seq
    * @param mapping
+   * @param dataset
+   *          - existing dataset. We check for sequences that look like the CDS
+   *          we are about to construct, if one exists already, then we will
+   *          just return that one.
    * @return CDS sequence (as a dataset sequence)
    */
-  static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping)
+  static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
+          AlignmentI dataset)
   {
     char[] seqChars = seq.getSequence();
     List<int[]> fromRanges = mapping.getMap().getFromRanges();
@@ -1879,6 +1917,40 @@ public class AlignmentUtils
     String mapFromId = mapping.getMappedFromId();
     String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
     SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
+    if (dataset != null)
+    {
+      SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
+      if (matches != null)
+      {
+        boolean matched = false;
+        for (SequenceI mtch : matches)
+        {
+          if (mtch.getStart() != newSeq.getStart())
+          {
+            continue;
+          }
+          if (mtch.getEnd() != newSeq.getEnd())
+          {
+            continue;
+          }
+          if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
+          {
+            continue;
+          }
+          if (!matched)
+          {
+            matched = true;
+            newSeq = mtch;
+          }
+          else
+          {
+            System.err
+                    .println("JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
+                            + mtch.toString());
+          }
+        }
+      }
+    }
     // newSeq.setDescription(mapFromId);
 
     return newSeq;
@@ -2568,7 +2640,7 @@ public class AlignmentUtils
   {
     AlignmentI copy = new Alignment(new Alignment(seqs));
     copy.setDataset(dataset);
-
+    boolean isProtein = !copy.isNucleotide();
     SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
     if (xrefs != null)
     {
@@ -2579,7 +2651,8 @@ public class AlignmentUtils
         {
           for (DBRefEntry dbref : dbrefs)
           {
-            if (dbref.getMap() == null || dbref.getMap().getTo() == null)
+            if (dbref.getMap() == null || dbref.getMap().getTo() == null
+                    || dbref.getMap().getTo().isProtein() != isProtein)
             {
               continue;
             }