X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Futil%2FDBRefUtils.java;h=f414a9c4c87d0b0336b88b408821633e17585c8d;hb=b9940c76e03cfc61550c94e5865019f5d6231daf;hp=405f6e6f520601ebd8785c2d26be1e5081475347;hpb=34e29960cefd8e743de8734ca0ea2778bab12a40;p=jalview.git

diff --git a/src/jalview/util/DBRefUtils.java b/src/jalview/util/DBRefUtils.java
index 405f6e6..f414a9c 100755
--- a/src/jalview/util/DBRefUtils.java
+++ b/src/jalview/util/DBRefUtils.java
@@ -26,6 +26,7 @@ import jalview.datamodel.PDBEntry;
 import jalview.datamodel.SequenceI;
 
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.HashMap;
 import java.util.HashSet;
 import java.util.Hashtable;
@@ -608,4 +609,119 @@ public class DBRefUtils
     return matches;
   }
 
+  /**
+   * promote direct database references to primary for nucleotide or protein
+   * sequences if they have an appropriate primary ref
+   * <table>
+   * <tr>
+   * <td>Seq Type</td>
+   * <td>Primary DB</td>
+   * <td>Direct which will be promoted</td>
+   * </tr>
+   * <tr>
+   * <td>peptides</td>
+   * <td>Ensembl</td>
+   * <td>Uniprot</td>
+   * </tr>
+   * <tr>
+   * <td>peptides</td>
+   * <td>Ensembl</td>
+   * <td>Uniprot</td>
+   * </tr>
+   * <tr>
+   * <td>dna</td>
+   * <td>Ensembl</td>
+   * <td>ENA</td>
+   * </tr>
+   * </table>
+   * 
+   * @param sequence
+   */
+  public static void ensurePrimaries(SequenceI sequence)
+  {
+    List<DBRefEntry> pr = sequence.getPrimaryDBRefs();
+    if (pr.size() == 0)
+    {
+      // nothing to do
+      return;
+    }
+    List<DBRefEntry> selfs = new ArrayList<DBRefEntry>();
+    selfs.addAll(Arrays.asList(selectDbRefs(!sequence.isProtein(),
+            sequence.getDBRefs())));
+
+    // filter non-primary refs
+    for (DBRefEntry p : pr)
+    {
+      while (selfs.contains(p))
+      {
+        selfs.remove(p);
+      }
+    }
+    List<DBRefEntry> toPromote = new ArrayList<DBRefEntry>();
+
+    for (DBRefEntry p : pr)
+    {
+      List<String> promType = new ArrayList<String>();
+      if (sequence.isProtein())
+      {
+        switch (getCanonicalName(p.getSource()))
+        {
+        case DBRefSource.UNIPROT:
+          // case DBRefSource.UNIPROTKB:
+          // case DBRefSource.UP_NAME:
+          // search for and promote ensembl
+          promType.add(DBRefSource.ENSEMBL);
+          break;
+        case DBRefSource.ENSEMBL:
+          // search for and promote Uniprot
+          promType.add(DBRefSource.UNIPROT);
+          break;
+        }
+      }
+      else
+      {
+        // TODO: promote transcript refs
+      }
+
+      // collate candidates and promote them
+      DBRefEntry[] candidates = selectRefs(
+              selfs.toArray(new DBRefEntry[0]),
+              promType.toArray(new String[0]));
+      if (candidates != null)
+      {
+        for (DBRefEntry cand : candidates)
+        {
+          if (cand.hasMap())
+          {
+            if (cand.getMap().getTo() != null
+                    && cand.getMap().getTo() != sequence)
+            {
+              // can't promote refs with mappings to other sequences
+              continue;
+            }
+            if (cand.getMap().getMap().getFromLowest() != sequence
+                    .getStart()
+                    && cand.getMap().getMap().getFromHighest() != sequence
+                            .getEnd())
+            {
+              // can't promote refs with mappings from a region of this sequence
+              // - eg CDS
+              continue;
+            }
+          }
+          // and promote
+          cand.setVersion(p.getVersion() + " (promoted)");
+          selfs.remove(cand);
+          toPromote.add(cand);
+          if (!cand.isPrimaryCandidate())
+          {
+            System.out.println("Warning: Couldn't promote dbref "
+                    + cand.toString() + " for sequence "
+                    + sequence.toString());
+          }
+        }
+      }
+    }
+  }
+
 }