Merge branch 'develop' into alpha/JAL-3362_Jalview_212_alpha

[jalview.git] / src / jalview / io / StockholmFile.java

diff --git a/src/jalview/io/StockholmFile.java b/src/jalview/io/StockholmFile.java
index f7a2df2..344c4e1 100644 (file)
--- a/src/jalview/io/StockholmFile.java
+++ b/src/jalview/io/StockholmFile.java
@@ -28,12 +28,14 @@ import jalview.datamodel.AlignmentAnnotation;
 import jalview.datamodel.AlignmentI;
 import jalview.datamodel.Annotation;
 import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.DBRefSource;
 import jalview.datamodel.Mapping;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
 import jalview.schemes.ResidueProperties;
 import jalview.util.Comparison;
+import jalview.util.DBRefUtils;
 import jalview.util.Format;
 import jalview.util.MessageManager;
 
@@ -337,17 +339,14 @@ public class StockholmFile extends AlignFile
 
           if (accAnnotations != null && accAnnotations.containsKey("AC"))
           {
-            if (dbsource != null)
+            String dbr = (String) accAnnotations.get("AC");
+            if (dbr != null)
             {
-              String dbr = (String) accAnnotations.get("AC");
-              if (dbr != null)
-              {
-                // we could get very clever here - but for now - just try to
-                // guess accession type from source of alignment plus structure
-                // of accession
-                guessDatabaseFor(seqO, dbr, dbsource);
-
-              }
+              // we could get very clever here - but for now - just try to
+              // guess accession type from type of sequence, source of alignment plus
+              // structure
+              // of accession
+              guessDatabaseFor(seqO, dbr, dbsource);
             }
             // else - do what ? add the data anyway and prompt the user to
             // specify what references these are ?
@@ -532,6 +531,9 @@ public class StockholmFile extends AlignFile
               treeName = an.stringMatched(2);
               treeString = new StringBuffer();
             }
+            // TODO: JAL-3532 - this is where GF comments and database references are lost
+            // suggest overriding this method for Stockholm files to catch and properly
+            // process CC, DR etc into multivalued properties
             setAlignmentProperty(an.stringMatched(1), an.stringMatched(2));
           }
         }
@@ -761,6 +763,12 @@ public class StockholmFile extends AlignFile
         st = -1;
       }
     }
+    if (dbsource == null)
+    {
+      // make up an origin based on whether the sequence looks like it is nucleotide
+      // or protein
+      dbsource = (seqO.isProtein()) ? "PFAM" : "RFAM";
+    }
     if (dbsource.equals("PFAM"))
     {
       seqdb = "UNIPROT";
@@ -940,6 +948,11 @@ public class StockholmFile extends AlignFile
     return annot;
   }
 
+  private String dbref_to_ac_record(DBRefEntry ref)
+  {
+    return ref.getSource().toString() + " ; "
+            + ref.getAccessionId().toString();
+  }
   @Override
   public String print(SequenceI[] s, boolean jvSuffix)
   {
@@ -952,8 +965,10 @@ public class StockholmFile extends AlignFile
     int maxid = 0;
     int in = 0;
     Hashtable dataRef = null;
+    boolean isAA = s[in].isProtein();
     while ((in < s.length) && (s[in] != null))
     {
+
       String tmp = printId(s[in], jvSuffix);
       max = Math.max(max, s[in].getLength());
 
@@ -963,17 +978,33 @@ public class StockholmFile extends AlignFile
       }
       if (s[in].getDBRefs() != null)
       {
-        for (int idb = 0; idb < s[in].getDBRefs().length; idb++)
+        if (dataRef == null)
+        {
+          dataRef = new Hashtable();
+        }
+        List<DBRefEntry> primrefs = s[in].getPrimaryDBRefs();
+        if (primrefs.size() >= 1)
         {
-          if (dataRef == null)
+          dataRef.put(tmp, dbref_to_ac_record(primrefs.get(0)));
+        }
+        else
+        {
+          for (int idb = 0; idb < s[in].getDBRefs().length; idb++)
           {
-            dataRef = new Hashtable();
+            DBRefEntry dbref = s[in].getDBRefs()[idb];
+            dataRef.put(tmp, dbref_to_ac_record(dbref));
+            // if we put in a uniprot or EMBL record then we're done:
+            if (isAA && DBRefSource.UNIPROT
+                    .equals(DBRefUtils.getCanonicalName(dbref.getSource())))
+            {
+              break;
+            }
+            if (!isAA && DBRefSource.EMBL
+                    .equals(DBRefUtils.getCanonicalName(dbref.getSource())))
+            {
+              break;
+            }
           }
-
-          String datAs1 = s[in].getDBRefs()[idb].getSource().toString()
-                  + " ; "
-                  + s[in].getDBRefs()[idb].getAccessionId().toString();
-          dataRef.put(tmp, datAs1);
         }
       }
       in++;
@@ -1006,7 +1037,8 @@ public class StockholmFile extends AlignFile
         String type = (String) dataRef.remove(idd);
         out.append(new Format("%-" + (maxid - 2) + "s")
                 .form("#=GS " + idd.toString() + " "));
-        if (type.contains("PFAM") || type.contains("RFAM"))
+        if (isAA && type.contains("UNIPROT")
+                || (!isAA && type.contains("EMBL")))
         {
 
           out.append(" AC " + type.substring(type.indexOf(";") + 1));