From: Jim Procter <jprocter@issues.jalview.org>
Date: Tue, 18 Feb 2020 18:38:12 +0000 (+0000)
Subject: JAL-3529 attempt to guess database for a sequence’s AC record based on the sequence... 
X-Git-Tag: Develop-2_11_2_0-d20201215~35^2~2
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=90cf1b4dbeb37554fd4b654c1da523f5791383e4;p=jalview.git

JAL-3529 attempt to guess database for a sequence’s AC record based on the sequence type if the alignment had no PFAM or RFAM accession
---

diff --git a/src/jalview/io/StockholmFile.java b/src/jalview/io/StockholmFile.java
index b9f1274..4697262 100644
--- a/src/jalview/io/StockholmFile.java
+++ b/src/jalview/io/StockholmFile.java
@@ -334,17 +334,14 @@ 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 ?
@@ -760,6 +757,12 @@ 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";
@@ -952,9 +955,10 @@ 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))
     {
-      boolean isAA = s[in].isProtein();
+
       String tmp = printId(s[in], jvSuffix);
       max = Math.max(max, s[in].getLength());
 
@@ -1023,7 +1027,8 @@ 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));