JAL-2920 use HGVS notation for protein variant feature

[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 fd68c12..11fff9d 100644 (file)
--- a/src/jalview/ws/dbsources/Uniprot.java
+++ b/src/jalview/ws/dbsources/Uniprot.java
@@ -20,6 +20,7 @@
  */
 package jalview.ws.dbsources;
 
+import jalview.bin.Cache;
 import jalview.datamodel.Alignment;
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.DBRefEntry;
@@ -31,6 +32,8 @@ import jalview.datamodel.SequenceI;
 import jalview.datamodel.xdb.uniprot.UniprotEntry;
 import jalview.datamodel.xdb.uniprot.UniprotFeature;
 import jalview.datamodel.xdb.uniprot.UniprotFile;
+import jalview.schemes.ResidueProperties;
+import jalview.util.StringUtils;
 import jalview.ws.seqfetcher.DbSourceProxyImpl;
 
 import java.io.InputStream;
@@ -52,6 +55,8 @@ import com.stevesoft.pat.Regex;
  */
 public class Uniprot extends DbSourceProxyImpl
 {
+  private static final String DEFAULT_UNIPROT_DOMAIN = "https://www.uniprot.org";
+
   private static final String BAR_DELIMITER = "|";
 
   /*
@@ -67,6 +72,11 @@ public class Uniprot extends DbSourceProxyImpl
     super();
   }
 
+  private String getDomain()
+  {
+    return Cache.getDefault("UNIPROT_DOMAIN", DEFAULT_UNIPROT_DOMAIN);
+  }
+
   /*
    * (non-Javadoc)
    * 
@@ -163,7 +173,7 @@ public class Uniprot extends DbSourceProxyImpl
               "(UNIPROT\\|?|UNIPROT_|UNIREF\\d+_|UNIREF\\d+\\|?)", "");
       AlignmentI al = null;
 
-      String downloadstring = "http://www.uniprot.org/uniprot/" + queries
+      String downloadstring = getDomain() + "/uniprot/" + queries
               + ".xml";
       URL url = null;
       URLConnection urlconn = null;
@@ -209,7 +219,7 @@ public class Uniprot extends DbSourceProxyImpl
     sequence.setDescription(getUniprotEntryDescription(entry));
 
     final String dbVersion = getDbVersion();
-    ArrayList<DBRefEntry> dbRefs = new ArrayList<DBRefEntry>();
+    ArrayList<DBRefEntry> dbRefs = new ArrayList<>();
     for (String accessionId : entry.getAccession())
     {
       DBRefEntry dbRef = new DBRefEntry(DBRefSource.UNIPROT, dbVersion,
@@ -219,7 +229,7 @@ public class Uniprot extends DbSourceProxyImpl
       dbRefs.add(dbRef);
     }
 
-    Vector<PDBEntry> onlyPdbEntries = new Vector<PDBEntry>();
+    Vector<PDBEntry> onlyPdbEntries = new Vector<>();
     for (PDBEntry pdb : entry.getDbReference())
     {
       DBRefEntry dbr = new DBRefEntry();
@@ -270,7 +280,7 @@ public class Uniprot extends DbSourceProxyImpl
       for (UniprotFeature uf : entry.getFeature())
       {
         SequenceFeature copy = new SequenceFeature(uf.getType(),
-                uf.getDescription(), uf.getBegin(), uf.getEnd(), "Uniprot");
+                getDescription(uf), uf.getBegin(), uf.getEnd(), "Uniprot");
         copy.setStatus(uf.getStatus());
         sequence.addSequenceFeature(copy);
       }
@@ -283,6 +293,43 @@ public class Uniprot extends DbSourceProxyImpl
   }
 
   /**
+   * Constructs a feature description from the description and (optionally)
+   * original and variant fields of the Uniprot XML feature
+   * 
+   * @param uf
+   * @return
+   */
+  protected static String getDescription(UniprotFeature uf)
+  {
+    String orig = uf.getOriginal();
+    String var = uf.getVariation();
+    StringBuilder sb = new StringBuilder();
+
+    /*
+     * append variant in standard format if present
+     * e.g. p.Arg59Lys
+     */
+    if (orig != null && !orig.isEmpty() && var != null && !var.isEmpty())
+    {
+      sb.append("p.");
+      String orig3 = ResidueProperties.aa2Triplet.get(orig);
+      sb.append(orig3 == null ? orig : StringUtils.toSentenceCase(orig3));
+      sb.append(Integer.toString(uf.getPosition()));
+      String var3 = ResidueProperties.aa2Triplet.get(var);
+      sb.append(var3 == null ? var : StringUtils.toSentenceCase(var3));
+      sb.append(" ");
+    }
+
+    String description = uf.getDescription();
+    if (description != null)
+    {
+      sb.append(description);
+    }
+
+    return sb.toString();
+  }
+
+  /**
    * 
    * @param entry
    *          UniportEntry
@@ -310,23 +357,18 @@ public class Uniprot extends DbSourceProxyImpl
   /**
    *
    * @param entry
-   *          UniportEntry
+   *          UniprotEntry
    * @return The accession id(s) and name(s) delimited by '|'.
    */
   public static String getUniprotEntryId(UniprotEntry entry)
   {
     StringBuilder name = new StringBuilder(32);
-    // 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);
-      name.append(accessionId);
-    }
     for (String n : entry.getName())
     {
-      name.append(BAR_DELIMITER);
+      if (name.length() > 0)
+      {
+        name.append(BAR_DELIMITER);
+      }
       name.append(n);
     }
     return name.toString();