JAL-2154 JAL-2106 transfer primary refs to CDS for makeCDS

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

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index b57fbbf..cc80384 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;
@@ -1682,6 +1681,10 @@ public class AlignmentUtils
            * its dataset sequence to the dataset
            */
           cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping);
+          // 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.createDatasetSequence();
           cdsSeqs.add(cdsSeq);
           if (!dataset.getSequences().contains(cdsSeqDss))
@@ -1708,6 +1711,8 @@ public class AlignmentUtils
             mappings.add(cdsToProteinMapping);
           }
 
+          propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
+                  proteinProduct, aMapping);
           /*
            * add another mapping from original 'from' range to CDS
            */
@@ -1729,12 +1734,37 @@ 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);
@@ -1879,8 +1909,6 @@ public class AlignmentUtils
     SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
     // newSeq.setDescription(mapFromId);
 
-    propagateDBRefsToCDS(newSeq, seq, mapping);
-
     return newSeq;
   }
 
@@ -1894,11 +1922,12 @@ public class AlignmentUtils
    * @return list of DBRefEntrys added.
    */
   public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
-          SequenceI contig, Mapping mapping)
+          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())
@@ -1910,24 +1939,51 @@ public class AlignmentUtils
           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)
     {
-      Mapping cdsmap = cdsref.getMap();
+      // clone maplist and mapping
       MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
-      { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsmap.getMap()
+      { 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);
     }