JAL-845 implement alignment of protein to match cDNA alignment

[jalview.git] / test / jalview / datamodel / AlignmentTest.java

diff --git a/test/jalview/datamodel/AlignmentTest.java b/test/jalview/datamodel/AlignmentTest.java
index 8912155..3b3d926 100644 (file)
--- a/test/jalview/datamodel/AlignmentTest.java
+++ b/test/jalview/datamodel/AlignmentTest.java
@@ -60,11 +60,13 @@ public class AlignmentTest
    * Helper method to load an alignment and ensure dataset sequences are set up.
    * 
    * @param data
-   * @param format TODO
+   * @param format
+   *          TODO
    * @return
    * @throws IOException
    */
-  protected AlignmentI loadAlignment(final String data, String format) throws IOException
+  protected AlignmentI loadAlignment(final String data, String format)
+          throws IOException
   {
     Alignment a = new FormatAdapter().readFile(data,
             AppletFormatAdapter.PASTE, format);
@@ -83,8 +85,7 @@ public class AlignmentTest
     int i = 0;
     for (AlignmentAnnotation ann : al.getAlignmentAnnotation())
     {
-      ann.setCalcId("CalcIdFor"
-              + al.getSequenceAt(i).getName());
+      ann.setCalcId("CalcIdFor" + al.getSequenceAt(i).getName());
       i++;
     }
   }
@@ -104,6 +105,35 @@ public class AlignmentTest
     assertFalse(iter.hasNext());
   }
 
+  @Test
+  public void testDeleteAllAnnotations_includingAutocalculated()
+  {
+    AlignmentAnnotation aa = new AlignmentAnnotation("Consensus",
+            "Consensus", 0.5);
+    aa.autoCalculated = true;
+    al.addAnnotation(aa);
+    AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
+    assertEquals("Wrong number of annotations before deleting", 4,
+            anns.length);
+    al.deleteAllAnnotations(true);
+    assertEquals("Not all deleted", 0, al.getAlignmentAnnotation().length);
+  }
+
+  @Test
+  public void testDeleteAllAnnotations_excludingAutocalculated()
+  {
+    AlignmentAnnotation aa = new AlignmentAnnotation("Consensus",
+            "Consensus", 0.5);
+    aa.autoCalculated = true;
+    al.addAnnotation(aa);
+    AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
+    assertEquals("Wrong number of annotations before deleting", 4,
+            anns.length);
+    al.deleteAllAnnotations(false);
+    assertEquals("Not just one annotation left", 1,
+            al.getAlignmentAnnotation().length);
+  }
+
   /**
    * Tests for realigning as per a supplied alignment: Dna as Dna.
    * 
@@ -133,7 +163,7 @@ public class AlignmentTest
     acf.addMap(al2.getSequenceAt(1), al1.getSequenceAt(1), ml);
     al1.addCodonFrame(acf);
 
-    al2.alignAs(al1);
+    ((Alignment) al2).alignAs(al1, false, true);
     assertEquals("GC-TC--GUC-GTA-CT", al2.getSequenceAt(0)
             .getSequenceAsString());
     assertEquals("-GG-GTC--AGG---CAGT", al2.getSequenceAt(1)
@@ -153,7 +183,7 @@ public class AlignmentTest
     String before0 = al2.getSequenceAt(0).getSequenceAsString();
     String before1 = al2.getSequenceAt(1).getSequenceAsString();
 
-    al2.alignAs(al1);
+    ((Alignment) al2).alignAs(al1, false, true);
     assertEquals(before0, al2.getSequenceAt(0).getSequenceAsString());
     assertEquals(before1, al2.getSequenceAt(1).getSequenceAsString());
   }
@@ -179,10 +209,69 @@ public class AlignmentTest
     acf.addMap(al1.getSequenceAt(1), al2.getSequenceAt(1), ml);
     al2.addCodonFrame(acf);
 
-    al1.alignAs(al2);
+    /*
+     * Realign DNA; currently keeping existing gaps in introns only
+     */
+    ((Alignment) al1).alignAs(al2, false, true);
     assertEquals("ACG---GCUCCA------ACT", al1.getSequenceAt(0)
             .getSequenceAsString());
-    assertEquals("---CGT---TAACGA---AGT---", al1.getSequenceAt(1)
+    assertEquals("---CGT---TAACGA---AGT", al1.getSequenceAt(1)
+            .getSequenceAsString());
+  }
+
+  /**
+   * Test aligning dna as per protein alignment, for the case where there are
+   * introns (i.e. some dna sites have no mapping from a peptide).
+   * 
+   * @throws IOException
+   */
+  @Test
+  public void testAlignAs_dnaAsProtein_withIntrons() throws IOException
+  {
+    /*
+     * Load alignments and add mappings for cDNA to protein
+     */
+    String dna1 = "A-Aa-gG-GCC-cT-TT";
+    String dna2 = "c--CCGgg-TT--T-AA-A";
+    AlignmentI al1 = loadAlignment(">Seq1\n" + dna1 + "\n>Seq2\n" + dna2
+            + "\n", "FASTA");
+    AlignmentI al2 = loadAlignment(">Seq1\n-P--YK\n>Seq2\nG-T--F\n",
+            "FASTA");
+    AlignedCodonFrame acf = new AlignedCodonFrame();
+    // Seq1 has intron at dna positions 3,4,9 so splice is AAG GCC TTT
+    // Seq2 has intron at dna positions 1,5,6 so splice is CCG TTT AAA
+    MapList ml1 = new MapList(new int[]
+    { 1, 2, 5, 8, 10, 12 }, new int[]
+    { 1, 3 }, 3, 1);
+    acf.addMap(al1.getSequenceAt(0), al2.getSequenceAt(0), ml1);
+    MapList ml2 = new MapList(new int[]
+    { 2, 4, 7, 12 }, new int[]
+    { 1, 3 }, 3, 1);
+    acf.addMap(al1.getSequenceAt(1), al2.getSequenceAt(1), ml2);
+    al2.addCodonFrame(acf);
+
+    /*
+     * Align ignoring gaps in dna introns and exons
+     */
+    ((Alignment) al1).alignAs(al2, false, false);
+    assertEquals("---AAagG------GCCcTTT", al1.getSequenceAt(0)
+            .getSequenceAsString());
+    // note 1 gap in protein corresponds to 'gg-' in DNA (3 positions)
+    assertEquals("cCCGgg-TTT------AAA", al1.getSequenceAt(1)
+            .getSequenceAsString());
+
+    /*
+     * Reset and realign, preserving gaps in dna introns and exons
+     */
+    al1.getSequenceAt(0).setSequence(dna1);
+    al1.getSequenceAt(1).setSequence(dna2);
+    ((Alignment) al1).alignAs(al2, true, true);
+    // String dna1 = "A-Aa-gG-GCC-cT-TT";
+    // String dna2 = "c--CCGgg-TT--T-AA-A";
+    // assumption: we include 'the greater of' protein/dna gap lengths, not both
+    assertEquals("---A-Aa-gG------GCC-cT-TT", al1.getSequenceAt(0)
+            .getSequenceAsString());
+    assertEquals("c--CCGgg-TT--T------AA-A", al1.getSequenceAt(1)
             .getSequenceAsString());
   }
 }