From 4b843c60054b25d20a745f2f015d7168037887df Mon Sep 17 00:00:00 2001
From: BobHanson <hansonr@stolaf.edu>
Date: Tue, 2 Jun 2020 07:27:43 -0500
Subject: [PATCH] JAL-3446 from JAL-3253-applet StockholmFile updated to
 26b115b (gmungoc)

---
 src/jalview/io/StockholmFile.java |   52 +++++++++++++++++++++++++++----------
 1 file changed, 39 insertions(+), 13 deletions(-)

diff --git a/src/jalview/io/StockholmFile.java b/src/jalview/io/StockholmFile.java
index 92f73f4..3d07e5a 100644
--- 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;
 import jalview.util.Platform;
@@ -421,17 +423,14 @@ public class StockholmFile extends AlignFile
 
           if (accAnnotations != null && accAnnotations.containsKey("AC"))
           {
-            if (dbsource != 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
+              // 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 ?
@@ -616,6 +615,9 @@ 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));
           }
         }
@@ -844,6 +846,12 @@ 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";
@@ -1020,6 +1028,12 @@ 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)
   {
@@ -1034,6 +1048,7 @@ public class StockholmFile extends AlignFile
     int slen = s.length;
     SequenceI seq;
     Hashtable<String, String> dataRef = null;
+    boolean isAA = s[in].isProtein();
     while ((in < slen) && ((seq = s[in]) != null))
     {
       String tmp = printId(seq, jvSuffix);
@@ -1051,14 +1066,24 @@ public class StockholmFile extends AlignFile
         {
           dataRef = new Hashtable<>();
         }
-        for (int idb = 0; idb < ndb; idb++)
+        List<DBRefEntry> primrefs = seq.getPrimaryDBRefs();
+        if (primrefs.size() >= 1)
         {
-
-          DBRefEntry ref = seqrefs.get(idb);
-          String datAs1 = ref.getSource().toString()
-                  + " ; "
-                  + ref.getAccessionId().toString();
-          dataRef.put(tmp, datAs1);
+          dataRef.put(tmp, dbref_to_ac_record(primrefs.get(0)));
+        }
+        else
+        {
+          for (int idb = 0; idb < ndb; idb++)
+          {
+            DBRefEntry dbref = seqrefs.get(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 : DBRefSource.EMBL)
+                    .equals(DBRefUtils.getCanonicalName(dbref.getSource())))
+            {
+              break;
+            }
+          }
         }
       }
       in++;
@@ -1091,7 +1116,8 @@ public class StockholmFile extends AlignFile
         String type = 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));
-- 
1.7.10.2