JAL-2446 merged to spike branch

[jalview.git] / src / jalview / ws / dbsources / Uniprot.java

diff --git a/src/jalview/ws/dbsources/Uniprot.java b/src/jalview/ws/dbsources/Uniprot.java
index 843828b..7261cba 100644 (file)
--- a/src/jalview/ws/dbsources/Uniprot.java
+++ b/src/jalview/ws/dbsources/Uniprot.java
@@ -28,18 +28,20 @@ import jalview.datamodel.PDBEntry;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
-import jalview.datamodel.UniprotEntry;
-import jalview.datamodel.UniprotFile;
+import jalview.datamodel.xdb.uniprot.UniprotEntry;
+import jalview.datamodel.xdb.uniprot.UniprotFeature;
+import jalview.datamodel.xdb.uniprot.UniprotFile;
 import jalview.ws.ebi.EBIFetchClient;
-import jalview.ws.seqfetcher.DbSourceProxy;
 import jalview.ws.seqfetcher.DbSourceProxyImpl;
 
 import java.io.File;
 import java.io.FileReader;
 import java.io.Reader;
+import java.net.URL;
 import java.util.ArrayList;
 import java.util.Vector;
 
+import org.exolab.castor.mapping.Mapping;
 import org.exolab.castor.xml.Unmarshaller;
 
 import com.stevesoft.pat.Regex;
@@ -48,12 +50,14 @@ import com.stevesoft.pat.Regex;
  * @author JimP
  * 
  */
-public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
+public class Uniprot extends DbSourceProxyImpl
 {
-
   private static final String BAR_DELIMITER = "|";
 
-  private static org.exolab.castor.mapping.Mapping map;
+  /*
+   * Castor mapping loaded from uniprot_mapping.xml
+   */
+  private static Mapping map;
 
   /**
    * Constructor
@@ -61,8 +65,6 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
   public Uniprot()
   {
     super();
-    addDbSourceProperty(DBRefSource.SEQDB, DBRefSource.SEQDB);
-    addDbSourceProperty(DBRefSource.PROTSEQDB);
   }
 
   /*
@@ -125,9 +127,8 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
       if (map == null)
       {
         // 1. Load the mapping information from the file
-        map = new org.exolab.castor.mapping.Mapping(uni.getClass()
-                .getClassLoader());
-        java.net.URL url = getClass().getResource("/uniprot_mapping.xml");
+        map = new Mapping(uni.getClass().getClassLoader());
+        URL url = getClass().getResource("/uniprot_mapping.xml");
         map.loadMapping(url);
       }
 
@@ -165,7 +166,7 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
       // uniprotxml parameter required since december 2007
       // uniprotkb dbname changed introduced december 2008
       File file = ebi.fetchDataAsFile("uniprotkb:" + queries, "uniprotxml",
-              null);
+              "xml");
       Vector<UniprotEntry> entries = getUniprotEntries(new FileReader(file));
 
       if (entries != null)
@@ -193,7 +194,8 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
    *          UniprotEntry
    * @return SequenceI instance created from the UniprotEntry instance
    */
-  public SequenceI uniprotEntryToSequenceI(UniprotEntry entry){
+  public SequenceI uniprotEntryToSequenceI(UniprotEntry entry)
+  {
     String id = getUniprotEntryId(entry);
     SequenceI sequence = new Sequence(id, entry.getUniprotSequence()
             .getContent());
@@ -205,10 +207,10 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
     {
       DBRefEntry dbRef = new DBRefEntry(DBRefSource.UNIPROT, dbVersion,
               accessionId);
+
+      // mark dbRef as a primary reference for this sequence
       dbRefs.add(dbRef);
     }
-    sequence.setSourceDBRef((dbRefs != null && dbRefs.size() > 0) ? dbRefs
-            .get(0) : null);
 
     Vector<PDBEntry> onlyPdbEntries = new Vector<PDBEntry>();
     for (PDBEntry pdb : entry.getDbReference())
@@ -222,18 +224,54 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
       {
         onlyPdbEntries.addElement(pdb);
       }
+      if ("EMBL".equals(pdb.getType()))
+      {
+        // look for a CDS reference and add it, too.
+        String cdsId = (String) pdb.getProperty("protein sequence ID");
+        if (cdsId != null && cdsId.trim().length() > 0)
+        {
+          // remove version
+          String[] vrs = cdsId.split("\\.");
+          dbr = new DBRefEntry(DBRefSource.EMBLCDS, vrs.length > 1 ? vrs[1]
+                  : DBRefSource.UNIPROT + ":" + dbVersion, vrs[0]);
+          dbRefs.add(dbr);
+        }
+      }
+      if ("Ensembl".equals(pdb.getType()))
+      {
+        /*UniprotXML
+         * <dbReference type="Ensembl" id="ENST00000321556">
+        * <molecule id="Q9BXM7-1"/>
+        * <property type="protein sequence ID" value="ENSP00000364204"/>
+        * <property type="gene ID" value="ENSG00000158828"/>
+        * </dbReference> 
+         */
+        String cdsId = (String) pdb.getProperty("protein sequence ID");
+        if (cdsId != null && cdsId.trim().length() > 0)
+        {
+          dbr = new DBRefEntry(DBRefSource.ENSEMBL, DBRefSource.UNIPROT
+                  + ":" + dbVersion, cdsId.trim());
+          dbRefs.add(dbr);
+
+        }
+      }
     }
 
     sequence.setPDBId(onlyPdbEntries);
     if (entry.getFeature() != null)
     {
-      for (SequenceFeature sf : entry.getFeature())
+      for (UniprotFeature uf : entry.getFeature())
       {
-        sf.setFeatureGroup("Uniprot");
-        sequence.addSequenceFeature(sf);
+        SequenceFeature copy = new SequenceFeature(uf.getType(),
+                uf.getDescription(), uf.getBegin(), uf.getEnd(), "Uniprot");
+        copy.setStatus(uf.getStatus());
+        sequence.addSequenceFeature(copy);
       }
     }
-    sequence.setDBRefs(dbRefs.toArray(new DBRefEntry[0]));
+    for (DBRefEntry dbr : dbRefs)
+    {
+      sequence.addDBRef(dbr);
+    }
     return sequence;
   }
 
@@ -248,9 +286,15 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
     StringBuilder desc = new StringBuilder(32);
     if (entry.getProtein() != null && entry.getProtein().getName() != null)
     {
+      boolean first = true;
       for (String nm : entry.getProtein().getName())
       {
-        desc.append(nm).append(" ");
+        if (!first)
+        {
+          desc.append(" ");
+        }
+        first = false;
+        desc.append(nm);
       }
     }
     return desc.toString();
@@ -265,7 +309,9 @@ public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
   public static String getUniprotEntryId(UniprotEntry entry)
   {
     StringBuilder name = new StringBuilder(32);
-    name.append("UniProt/Swiss-Prot");
+    // name.append("UniProt/Swiss-Prot");
+    // use 'canonicalised' name for optimal id matching
+    name.append(DBRefSource.UNIPROT);
     for (String accessionId : entry.getAccession())
     {
       name.append(BAR_DELIMITER);