From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 29 Mar 2018 09:47:17 +0000 (+0100)
Subject: JAL-2907 bug fix redone on HMMER branch
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=fdf1d81fa8cd498d17b959427930e3fc0bc85249;p=jalview.git

JAL-2907 bug fix redone on HMMER branch
---

diff --git a/src/jalview/io/StockholmFile.java b/src/jalview/io/StockholmFile.java
index 09c5b25..58b171d 100644
--- a/src/jalview/io/StockholmFile.java
+++ b/src/jalview/io/StockholmFile.java
@@ -79,6 +79,8 @@ public class StockholmFile extends AlignFile
 {
   private static final String ANNOTATION = "annotation";
 
+  private static final char UNDERSCORE = '_';
+
   // private static final Regex OPEN_PAREN = new Regex("(<|\\[)", "(");
   // private static final Regex CLOSE_PAREN = new Regex("(>|\\])", ")");
 
@@ -111,14 +113,14 @@ public class StockholmFile extends AlignFile
     featureTypes.put("RF", "Reference Positions");
   }
 
-  AlignmentI al;
+  private AlignmentI al;
 
   public StockholmFile()
   {
   }
 
   /**
-   * Creates a new StockholmFile object for output.
+   * Creates a new StockholmFile object for output
    */
   public StockholmFile(AlignmentI al)
   {
@@ -903,6 +905,10 @@ public class StockholmFile extends AlignFile
     for (int i = 0; i < annots.length(); i++)
     {
       String pos = annots.substring(i, i + 1);
+      if (UNDERSCORE == pos.charAt(0))
+      {
+        pos = " ";
+      }
       Annotation ann;
       ann = new Annotation(pos, "", ' ', 0f); // 0f is 'valid' null - will not
       // be written out
@@ -1171,7 +1177,7 @@ public class StockholmFile extends AlignFile
    * @param ann
    * @param sequenceI
    */
-  private char getAnnotationCharacter(String key, int k, Annotation annot,
+  static char getAnnotationCharacter(String key, int k, Annotation annot,
           SequenceI sequenceI)
   {
     char seq = ' ';
@@ -1183,8 +1189,8 @@ public class StockholmFile extends AlignFile
     {
       if (annot == null)
       {
-        // sensible gap character
-        return ' ';
+        // Stockholm format requires underscore, not space
+        return UNDERSCORE;
       }
       else
       {
diff --git a/test/jalview/io/StockholmFileTest.java b/test/jalview/io/StockholmFileTest.java
index 14050e8..7cc5258 100644
--- a/test/jalview/io/StockholmFileTest.java
+++ b/test/jalview/io/StockholmFileTest.java
@@ -37,6 +37,7 @@ import jalview.datamodel.SequenceI;
 import jalview.gui.JvOptionPane;
 
 import java.io.File;
+import java.io.IOException;
 import java.util.Arrays;
 import java.util.BitSet;
 import java.util.HashMap;
@@ -822,4 +823,109 @@ public class StockholmFileTest
     //@formatter:on
     assertEquals(expected, output);
   }
+
+  @Test(groups = "Functional")
+  public void testOutputCharacter()
+  {
+    SequenceI seq = new Sequence("seq", "abc--def-");
+
+    Annotation[] ann = new Annotation[8];
+    ann[1] = new Annotation("Z", "desc", 'E', 1f);
+    ann[2] = new Annotation("Q", "desc", ' ', 1f);
+    ann[4] = new Annotation("", "desc", 'E', 1f);
+    ann[6] = new Annotation("ZH", "desc", 'E', 1f);
+
+    /*
+     * null annotation in column (not Secondary Structure annotation)
+     * should answer sequence character, or '-' if null sequence
+     */
+    assertEquals('-',
+            StockholmFile.getAnnotationCharacter("RF", 0, ann[0], null));
+    assertEquals('d',
+            StockholmFile.getAnnotationCharacter("RF", 5, ann[5], seq));
+    assertEquals('-',
+            StockholmFile.getAnnotationCharacter("RF", 8, null, seq));
+
+    /*
+     * null annotation in column (SS annotation) should answer underscore
+     */
+    assertEquals('_',
+            StockholmFile.getAnnotationCharacter("SS", 0, ann[0], seq));
+
+    /*
+     * SS secondary structure symbol
+     */
+    assertEquals('E',
+            StockholmFile.getAnnotationCharacter("SS", 1, ann[1], seq));
+
+    /*
+     * no SS symbol, use label instead 
+     */
+    assertEquals('Q',
+            StockholmFile.getAnnotationCharacter("SS", 2, ann[2], seq));
+
+    /*
+     * SS with 2 character label - second character overrides SS symbol 
+     */
+    assertEquals('H',
+            StockholmFile.getAnnotationCharacter("SS", 6, ann[6], seq));
+
+    /*
+     * empty display character, not SS - answers '.'
+     */
+    assertEquals('.',
+            StockholmFile.getAnnotationCharacter("RF", 4, ann[4], seq));
+  }
+
+  /**
+   * Test to verify that gaps are input/output as underscore in STO annotation
+   * 
+   * @throws IOException
+   */
+  @Test(groups = "Functional")
+  public void testRoundtripWithGaps() throws IOException
+  {
+    /*
+     * small extract from RF00031_folded.stk
+     */
+    // @formatter:off
+    String stoData = 
+            "# STOCKHOLM 1.0\n" +
+            "#=GR B.taurus.4 SS .._((.))_\n" +
+            "B.taurus.4         AC.UGCGU.\n" +
+            "#=GR B.taurus.5 SS ..((_._))\n" +
+            "B.taurus.5         ACUU.G.CG\n" +
+        "//\n";
+    // @formatter:on
+    StockholmFile parser = new StockholmFile(stoData, DataSourceType.PASTE);
+    SequenceI[] seqs = parser.getSeqsAsArray();
+    assertEquals(2, seqs.length);
+
+    /*
+     * B.taurus.4 has a trailing gap
+     * rendered as underscore in Stockholm annotation
+     */
+    assertEquals("AC.UGCGU.", seqs[0].getSequenceAsString());
+    AlignmentAnnotation[] anns = seqs[0].getAnnotation();
+    assertEquals(1, anns.length);
+    AlignmentAnnotation taurus4SS = anns[0];
+    assertEquals(9, taurus4SS.annotations.length);
+    assertEquals(" .", taurus4SS.annotations[0].displayCharacter);
+    assertNull(taurus4SS.annotations[2]); // gapped position
+    assertNull(taurus4SS.annotations[8]); // gapped position
+    assertEquals('(', taurus4SS.annotations[3].secondaryStructure);
+    assertEquals("(", taurus4SS.annotations[3].displayCharacter);
+    assertEquals(')', taurus4SS.annotations[7].secondaryStructure);
+
+    /*
+     * output as Stockholm and verify it matches the original input
+     * (gaps output as underscore in annotation lines)
+     * note: roundtrip test works with the input lines ordered as above;
+     * can also parse in other orders, but then input doesn't match output
+     */
+    AlignmentFileWriterI afile = FileFormat.Stockholm
+            .getWriter(new Alignment(seqs));
+    String output = afile.print(seqs, false);
+    assertEquals(stoData, output);
+  }
 }