JAL-3061

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

diff --git a/src/jalview/io/StockholmFile.java b/src/jalview/io/StockholmFile.java
index f5b5177..0e73af1 100644 (file)
--- a/src/jalview/io/StockholmFile.java
+++ b/src/jalview/io/StockholmFile.java
@@ -83,6 +83,14 @@ public class StockholmFile extends AlignFile
   public static final Regex DETECT_BRACKETS = new Regex(
           "(<|>|\\[|\\]|\\(|\\)|\\{|\\})");
 
+  // WUSS extended symbols. Avoid ambiguity with protein SS annotations by using NOT_RNASS first.
+  public static final String RNASS_BRACKETS = "<>[](){}AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz";
+
+  // use the following regex to decide an annotations (whole) line is NOT an RNA
+  // SS (it contains only E,H,e,h and other non-brace/non-alpha chars)
+  private static final Regex NOT_RNASS = new Regex(
+          "^[^<>[\\](){}A-DF-Za-df-z]*$");
+
   StringBuffer out; // output buffer
 
   AlignmentI al;
@@ -197,7 +205,7 @@ public class StockholmFile extends AlignFile
     String version;
     // String id;
     Hashtable seqAnn = new Hashtable(); // Sequence related annotations
-    LinkedHashMap<String, String> seqs = new LinkedHashMap<String, String>();
+    LinkedHashMap<String, String> seqs = new LinkedHashMap<>();
     Regex p, r, rend, s, x;
     // Temporary line for processing RNA annotation
     // String RNAannot = "";
@@ -658,7 +666,7 @@ public class StockholmFile extends AlignFile
               strucAnn = new Hashtable();
             }
 
-            Vector<AlignmentAnnotation> newStruc = new Vector<AlignmentAnnotation>();
+            Vector<AlignmentAnnotation> newStruc = new Vector<>();
             parseAnnotationRow(newStruc, type, ns);
             for (AlignmentAnnotation alan : newStruc)
             {
@@ -710,7 +718,7 @@ public class StockholmFile extends AlignFile
   private void guessDatabaseFor(Sequence seqO, String dbr, String dbsource)
   {
     DBRefEntry dbrf = null;
-    List<DBRefEntry> dbrs = new ArrayList<DBRefEntry>();
+    List<DBRefEntry> dbrs = new ArrayList<>();
     String seqdb = "Unknown", sdbac = "" + dbr;
     int st = -1, en = -1, p;
     if ((st = sdbac.indexOf("/")) > -1)
@@ -824,9 +832,14 @@ public class StockholmFile extends AlignFile
     }
     boolean ss = false, posterior = false;
     type = id2type(type);
+
+    boolean isrnass = false;
     if (type.equalsIgnoreCase("secondary structure"))
     {
       ss = true;
+      isrnass = !NOT_RNASS.search(annots); // sorry about the double negative
+                                           // here (it's easier for dealing with
+                                           // other non-alpha-non-brace chars)
     }
     if (type.equalsIgnoreCase("posterior probability"))
     {
@@ -844,7 +857,7 @@ public class StockholmFile extends AlignFile
       {
         // if (" .-_".indexOf(pos) == -1)
         {
-          if (DETECT_BRACKETS.search(pos))
+          if (isrnass && RNASS_BRACKETS.indexOf(pos) >= 0)
           {
             ann.secondaryStructure = Rna.getRNASecStrucState(pos).charAt(0);
             ann.displayCharacter = "" + pos.charAt(0);
@@ -1114,22 +1127,36 @@ public class StockholmFile extends AlignFile
     String ch = (annot == null)
             ? ((sequenceI == null) ? "-"
                     : Character.toString(sequenceI.getCharAt(k)))
-            : annot.displayCharacter;
+            : (annot.displayCharacter == null
+                    ? String.valueOf(annot.secondaryStructure)
+                    : annot.displayCharacter);
+    if (ch == null)
+    {
+      ch = " ";
+    }
     if (key != null && key.equals("SS"))
     {
+      char ssannotchar = ' ';
+      boolean charset = false;
       if (annot == null)
       {
         // sensible gap character
-        return ' ';
+        ssannotchar = ' ';
+        charset = true;
       }
       else
       {
         // valid secondary structure AND no alternative label (e.g. ' B')
         if (annot.secondaryStructure > ' ' && ch.length() < 2)
         {
-          return annot.secondaryStructure;
+          ssannotchar = annot.secondaryStructure;
+          charset = true;
         }
       }
+      if (charset)
+      {
+        return (ssannotchar == ' ' && isrna) ? '.' : ssannotchar;
+      }
     }
 
     if (ch.length() == 0)
@@ -1144,7 +1171,9 @@ public class StockholmFile extends AlignFile
     {
       seq = ch.charAt(1);
     }
-    return seq;
+
+    return (seq == ' ' && key != null && key.equals("SS") && isrna) ? '.'
+            : seq;
   }
 
   public String print()