X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FAlignmentUtils.java;fp=src%2Fjalview%2Fanalysis%2FAlignmentUtils.java;h=228446bc7cdb8162d58d4e40b349141ebaf7c6ed;hb=e464da0ea20b0322846ce623598a16f2a1cba7ae;hp=9a46a919ee8a211003de54070d11fd42aad1b9e3;hpb=f708b46b35fe64dce97f5c8420104ba71cc21ff2;p=jalview.git

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index 9a46a91..228446b 100644
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -29,6 +29,7 @@ import jalview.datamodel.Alignment;
 import jalview.datamodel.AlignmentAnnotation;
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.GeneLociI;
 import jalview.datamodel.IncompleteCodonException;
 import jalview.datamodel.Mapping;
 import jalview.datamodel.Sequence;
@@ -394,7 +395,7 @@ public class AlignmentUtils
    * Answers true if the mappings include one between the given (dataset)
    * sequences.
    */
-  public static boolean mappingExists(List<AlignedCodonFrame> mappings,
+  protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
           SequenceI aaSeq, SequenceI cdnaSeq)
   {
     if (mappings != null)
@@ -1633,7 +1634,7 @@ public class AlignmentUtils
           AlignmentI dataset, SequenceI[] products)
   {
     if (dataset == null || dataset.getDataset() != null)
-  {
+    {
       throw new IllegalArgumentException(
               "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
     }
@@ -1645,10 +1646,10 @@ public class AlignmentUtils
     {
       productSeqs = new HashSet<SequenceI>();
       for (SequenceI seq : products)
-    {
+      {
         productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
                 .getDatasetSequence());
-    }
+      }
     }
 
     /*
@@ -1670,15 +1671,15 @@ public class AlignmentUtils
       List<AlignedCodonFrame> seqMappings = MappingUtils
               .findMappingsForSequence(dnaSeq, mappings);
       for (AlignedCodonFrame mapping : seqMappings)
-    {
+      {
         List<Mapping> mappingsFromSequence = mapping
                 .getMappingsFromSequence(dnaSeq);
 
         for (Mapping aMapping : mappingsFromSequence)
-      {
+        {
           MapList mapList = aMapping.getMap();
           if (mapList.getFromRatio() == 1)
-        {
+          {
             /*
              * not a dna-to-protein mapping (likely dna-to-cds)
              */
@@ -1704,15 +1705,15 @@ public class AlignmentUtils
           if (cdsSeq != null)
           {
             if (!foundSeqs.contains(cdsSeq))
-          {
+            {
               foundSeqs.add(cdsSeq);
               SequenceI derivedSequence = cdsSeq.deriveSequence();
               cdsSeqs.add(derivedSequence);
               if (!dataset.getSequences().contains(cdsSeq))
-            {
+              {
                 dataset.addSequence(cdsSeq);
+              }
             }
-          }
             continue;
           }
 
@@ -1763,7 +1764,7 @@ public class AlignmentUtils
            * add another mapping from original 'from' range to CDS
            */
           AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
-          MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
+          final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
                   cdsRange, 1, 1);
           dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
                   dnaToCdsMap);
@@ -1773,6 +1774,13 @@ public class AlignmentUtils
           }
 
           /*
+           * transfer dna chromosomal loci (if known) to the CDS
+           * sequence (via the mapping)
+           */
+          final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
+          transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
+
+          /*
            * add DBRef with mapping from protein to CDS
            * (this enables Get Cross-References from protein alignment)
            * This is tricky because we can't have two DBRefs with the
@@ -1795,13 +1803,11 @@ public class AlignmentUtils
              * create a cross-reference from CDS to the source sequence's
              * primary reference and vice versa
              */
-
             String source = primRef.getSource();
             String version = primRef.getVersion();
             DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
                     + version, primRef.getAccessionId());
-            cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap
-                    .getInverse())));
+            cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
             cdsSeqDss.addDBRef(cdsCrossRef);
 
             dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
@@ -1818,16 +1824,16 @@ public class AlignmentUtils
             proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
                     .getInverse()));
             proteinProduct.addDBRef(proteinToCdsRef);
-        }
+          }
 
           /*
            * transfer any features on dna that overlap the CDS
            */
           transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
                   SequenceOntologyI.CDS);
+        }
       }
     }
-    }
 
     AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
             .size()]));
@@ -1837,6 +1843,38 @@ public class AlignmentUtils
   }
 
   /**
+   * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
+   * toSeq, mediated by the given mapping between the sequences
+   * 
+   * @param fromSeq
+   * @param targetToFrom
+   *          Map
+   * @param targetSeq
+   */
+  protected static void transferGeneLoci(SequenceI fromSeq,
+          MapList targetToFrom, SequenceI targetSeq)
+  {
+    if (targetSeq.getGeneLoci() != null)
+    {
+      // already have - don't override
+      return;
+    }
+    GeneLociI fromLoci = fromSeq.getGeneLoci();
+    if (fromLoci == null)
+    {
+      return;
+    }
+
+    MapList newMap = targetToFrom.traverse(fromLoci.getMap());
+
+    if (newMap != null)
+    {
+      targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
+              fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
+    }
+  }
+
+  /**
    * A helper method that finds a CDS sequence in the alignment dataset that is
    * mapped to the given protein sequence, and either is, or has a mapping from,
    * the given dna sequence.
@@ -2004,19 +2042,19 @@ public class AlignmentUtils
   }
 
   /**
-   * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
+   * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
    * the given mapping.
    * 
    * @param cdsSeq
    * @param contig
+   * @param proteinProduct
    * @param mapping
-   * @return list of DBRefEntrys added.
+   * @return list of DBRefEntrys added
    */
-  public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
+  protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
           SequenceI contig, SequenceI proteinProduct, Mapping mapping)
   {
-
-    // gather direct refs from contig congrent with mapping
+    // gather direct refs from contig congruent with mapping
     List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
     HashSet<String> directSources = new HashSet<String>();
     if (contig.getDBRefs() != null)
@@ -2096,7 +2134,7 @@ public class AlignmentUtils
    *          subtypes in the Sequence Ontology)
    * @param omitting
    */
-  public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
+  protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
           MapList mapping, String select, String... omitting)
   {
     SequenceI copyTo = toSeq;
@@ -2234,7 +2272,7 @@ public class AlignmentUtils
    * @param dnaSeq
    * @return
    */
-  public static List<int[]> findCdsPositions(SequenceI dnaSeq)
+  protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
   {
     List<int[]> result = new ArrayList<int[]>();