Merge branch 'develop' into merge/develop_bug/JAL-2154projectMappings

[jalview.git] / src / jalview / ext / ensembl / EnsemblSeqProxy.java

diff --git a/src/jalview/ext/ensembl/EnsemblSeqProxy.java b/src/jalview/ext/ensembl/EnsemblSeqProxy.java
index 5f3f1c8..5fccedd 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblSeqProxy.java
+++ b/src/jalview/ext/ensembl/EnsemblSeqProxy.java
@@ -5,7 +5,6 @@ import jalview.analysis.Dna;
 import jalview.datamodel.Alignment;
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
-import jalview.datamodel.DBRefSource;
 import jalview.datamodel.Mapping;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
@@ -14,6 +13,7 @@ import jalview.io.FastaFile;
 import jalview.io.FileParse;
 import jalview.io.gff.SequenceOntologyFactory;
 import jalview.io.gff.SequenceOntologyI;
+import jalview.util.Comparison;
 import jalview.util.DBRefUtils;
 import jalview.util.MapList;
 
@@ -36,12 +36,6 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
 {
   private static final String ALLELES = "alleles";
 
-  private static final List<String> CROSS_REFERENCES = Arrays
-          .asList(new String[] { "CCDS", "Uniprot/SWISSPROT",
-              "Uniprot/SPTREMBL" });
-
-  protected static final String CONSEQUENCE_TYPE = "consequence_type";
-
   protected static final String PARENT = "Parent";
 
   protected static final String ID = "ID";
@@ -61,7 +55,7 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
     GENOMIC("genomic"),
 
     /**
-     * type=cdna to fetch dna including UTRs
+     * type=cdna to fetch coding dna including UTRs
      */
     CDNA("cdna"),
 
@@ -282,11 +276,10 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
       {
         // clunky: ensure Uniprot xref if we have one is on mapped sequence
         SequenceI ds = proteinSeq.getDatasetSequence();
-        ds.setSourceDBRef(proteinSeq.getSourceDBRef());
-
+        // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
         Mapping map = new Mapping(ds, mapList);
-        DBRefEntry dbr = new DBRefEntry(getDbSource(), getDbVersion(),
-                proteinSeq.getName(), map);
+        DBRefEntry dbr = new DBRefEntry(getDbSource(),
+                getEnsemblDataVersion(), proteinSeq.getName(), map);
         querySeq.getDatasetSequence().addDBRef(dbr);
         
         /*
@@ -315,41 +308,23 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
       seq = seq.getDatasetSequence();
     }
 
-    EnsemblXref xrefFetcher = new EnsemblXref(getDomain());
-    List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName(),
-            getCrossReferenceDatabases());
+    EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
+            getEnsemblDataVersion());
+    List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
     for (DBRefEntry xref : xrefs)
     {
       seq.addDBRef(xref);
-      /*
-       * Save any Uniprot xref to be the reference for SIFTS mapping
-       */
-      if (DBRefSource.UNIPROT.equals(xref.getSource()))
-      {
-        seq.setSourceDBRef(xref);
-      }
     }
 
     /*
      * and add a reference to itself
      */
-    DBRefEntry self = new DBRefEntry(getDbSource(), "0", seq.getName());
+    DBRefEntry self = new DBRefEntry(getDbSource(),
+            getEnsemblDataVersion(), seq.getName());
     seq.addDBRef(self);
   }
 
   /**
-   * Returns a list of database names to be used when fetching cross-references.
-   * Specifically, the names are used to filter data returned by the Ensembl
-   * xrefs REST service on the value in field 'dbname'.
-   * 
-   * @return
-   */
-  protected List<String> getCrossReferenceDatabases()
-  {
-    return CROSS_REFERENCES;
-  }
-
-  /**
    * Fetches sequences for the list of accession ids and adds them to the
    * alignment. Returns the extended (or created) alignment.
    * 
@@ -404,7 +379,9 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
         if (ids.contains(name)
                 || ids.contains(name.replace("ENSP", "ENST")))
         {
-          DBRefUtils.parseToDbRef(sq, DBRefSource.ENSEMBL, "0", name);
+          DBRefEntry dbref = DBRefUtils.parseToDbRef(sq, getDbSource(),
+                  getEnsemblDataVersion(), name);
+          sq.addDBRef(dbref);
         }
       }
       if (alignment == null)
@@ -635,6 +612,10 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
       SequenceFeature copy = new SequenceFeature(sf);
       copy.setBegin(Math.min(mappedRange[0], mappedRange[1]));
       copy.setEnd(Math.max(mappedRange[0], mappedRange[1]));
+      if (".".equals(copy.getFeatureGroup()))
+      {
+        copy.setFeatureGroup(getDbSource());
+      }
       targetSequence.addSequenceFeature(copy);
 
       /*
@@ -698,16 +679,21 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
     {
       complement.append(",");
     }
-    if ("HGMD_MUTATION".equalsIgnoreCase(allele))
+
+    /*
+     * some 'alleles' are actually descriptive terms 
+     * e.g. HGMD_MUTATION, PhenCode_variation
+     * - we don't want to 'reverse complement' these
+     */
+    if (!Comparison.isNucleotideSequence(allele, true))
     {
       complement.append(allele);
     }
     else
     {
-      char[] alleles = allele.toCharArray();
-      for (int i = alleles.length - 1; i >= 0; i--)
+      for (int i = allele.length() - 1; i >= 0; i--)
       {
-        complement.append(Dna.getComplement(alleles[i]));
+        complement.append(Dna.getComplement(allele.charAt(i)));
       }
     }
   }