Merge branch 'develop' into merge/develop_bug/JAL-2154projectMappings
[jalview.git] / src / jalview / analysis / AlignmentUtils.java
index d1cd5a3..ea330d8 100644 (file)
@@ -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;
@@ -1689,11 +1688,20 @@ 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);
           }
 
@@ -1705,7 +1713,8 @@ public class AlignmentUtils
           MapList cdsToProteinMap = new MapList(cdsRange, mapList.getToRanges(),
                   mapList.getFromRatio(), mapList.getToRatio());
           AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
-          cdsToProteinMapping.addMap(cdsSeq, proteinProduct, cdsToProteinMap);
+          cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
+                  cdsToProteinMap);
 
           /*
            * guard against duplicating the mapping if repeating this action
@@ -1715,23 +1724,8 @@ public class AlignmentUtils
             mappings.add(cdsToProteinMapping);
           }
 
-          /*
-           * copy protein's dbrefs to CDS sequence
-           * this enables Get Cross-References from CDS alignment
-           */
-          DBRefEntry[] proteinRefs = DBRefUtils.selectDbRefs(false,
-                  proteinProduct.getDBRefs());
-          if (proteinRefs != null)
-          {
-            for (DBRefEntry ref : proteinRefs)
-            {
-              DBRefEntry cdsToProteinRef = new DBRefEntry(ref);
-              cdsToProteinRef.setMap(new Mapping(proteinProduct,
-                      cdsToProteinMap));
-              cdsSeqDss.addDBRef(cdsToProteinRef);
-            }
-          }
-
+          propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
+                  proteinProduct, aMapping);
           /*
            * add another mapping from original 'from' range to CDS
            */
@@ -1739,7 +1733,7 @@ public class AlignmentUtils
           MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
                   cdsRange, 1,
                   1);
-          dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeq,
+          dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
                   dnaToCdsMap);
           if (!mappings.contains(dnaToCdsMapping))
           {
@@ -1753,12 +1747,37 @@ 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);
@@ -1864,9 +1883,14 @@ 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();
@@ -1893,7 +1917,7 @@ public class AlignmentUtils
         }
       }
     }
-
+    
     /*
      * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
      * else generate a sequence name
@@ -1901,12 +1925,124 @@ 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;
   }
 
   /**
+   * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
+   * the given mapping.
+   * 
+   * @param cdsSeq
+   * @param contig
+   * @param mapping
+   * @return list of DBRefEntrys added.
+   */
+  public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
+          SequenceI contig, SequenceI proteinProduct, Mapping mapping)
+  {
+
+    // gather direct refs from contig congrent with mapping
+    List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
+    HashSet<String> directSources = new HashSet<String>();
+    if (contig.getDBRefs() != null)
+    {
+      for (DBRefEntry dbr : contig.getDBRefs())
+      {
+        if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
+        {
+          MapList map = dbr.getMap().getMap();
+          // check if map is the CDS mapping
+          if (mapping.getMap().equals(map))
+          {
+            direct.add(dbr);
+            directSources.add(dbr.getSource());
+          }
+        }
+      }
+    }
+    DBRefEntry[] onSource = DBRefUtils.selectRefs(
+            proteinProduct.getDBRefs(),
+            directSources.toArray(new String[0]));
+    List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
+
+    // and generate appropriate mappings
+    for (DBRefEntry cdsref : direct)
+    {
+      // clone maplist and mapping
+      MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
+      { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsref.getMap().getMap()
+              .getToRanges(), 3, 1);
+      Mapping cdsmap = new Mapping(cdsref.getMap().getTo(), cdsref.getMap()
+              .getMap());
+
+      // create dbref
+      DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
+              cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
+                      cdsmap.getTo(), cdsposmap));
+
+      // and see if we can map to the protein product for this mapping.
+      // onSource is the filtered set of accessions on protein that we are
+      // tranferring, so we assume accession is the same.
+      if (cdsmap.getTo() == null && onSource != null)
+      {
+        List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
+                cdsref.getAccessionId());
+        if (sourceRefs != null)
+        {
+          for (DBRefEntry srcref : sourceRefs)
+          {
+            if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
+            {
+              // we have found a complementary dbref on the protein product, so
+              // update mapping's getTo
+              newref.getMap().setTo(proteinProduct);
+            }
+          }
+        }
+      }
+      cdsSeq.addDBRef(newref);
+      propagated.add(newref);
+    }
+    return propagated;
+  }
+
+  /**
    * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
    * feature start/end ranges, optionally omitting specified feature types.
    * Returns the number of features copied.
@@ -2513,7 +2649,7 @@ 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)
     {
@@ -2524,7 +2660,8 @@ 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;
             }