X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=test%2Fjalview%2Fio%2FScoreMatrixFileTest.java;h=99ec9a6bcd08d7791c59a6e7fbb905c07f5df490;hb=57738a1f3c19b1c3a00bd3ac5108f8cd0af32f99;hp=77b7282531e2fcc290d15b9c8faee6b2a5c03982;hpb=45e015aabe8f35a4a13be26e7630641ef8c94fbb;p=jalview.git

diff --git a/test/jalview/io/ScoreMatrixFileTest.java b/test/jalview/io/ScoreMatrixFileTest.java
index 77b7282..99ec9a6 100644
--- a/test/jalview/io/ScoreMatrixFileTest.java
+++ b/test/jalview/io/ScoreMatrixFileTest.java
@@ -3,19 +3,28 @@ package jalview.io;
 import static org.testng.Assert.assertEquals;
 import static org.testng.Assert.assertFalse;
 import static org.testng.Assert.assertNotNull;
+import static org.testng.Assert.assertNull;
 import static org.testng.Assert.assertTrue;
 import static org.testng.Assert.fail;
 
 import jalview.analysis.scoremodels.ScoreMatrix;
+import jalview.analysis.scoremodels.ScoreModels;
 
 import java.io.IOException;
 import java.net.MalformedURLException;
 
+import org.testng.annotations.AfterMethod;
 import org.testng.annotations.Test;
 
 public class ScoreMatrixFileTest
 {
 
+  @AfterMethod(alwaysRun = true)
+  public void tearDownAfterTest()
+  {
+    ScoreModels.getInstance().reset();
+  }
+
   /**
    * Test a successful parse of a (small) score matrix file
    * 
@@ -23,63 +32,52 @@ public class ScoreMatrixFileTest
    * @throws MalformedURLException
    */
   @Test(groups = "Functional")
-  public void testParse() throws MalformedURLException, IOException
+  public void testParseMatrix_ncbiMixedDelimiters()
+          throws MalformedURLException, IOException
   {
     /*
      * some messy but valid input data, with comma, space
      * or tab (or combinations) as score value delimiters
      * this example includes 'guide' symbols on score rows
      */
-    String data = "ScoreMatrix MyTest\n" + "ATU tx-\n"
-            + "A,1.1,1.2,1.3,1.4, 1.5, 1.6, 1.7\n"
-            + "T,2.1 2.2 2.3 2.4 2.5 2.6 2.7\n"
-            + "U\t3.1\t3.2\t3.3\t3.4\t3.5\t3.6\t3.7\n"
-            + " 4.1 ,4.2,\t,4.3 ,\t4.4\t, \4.5,4.6 4.7\n"
-            + "t, 5.1,5.3,5.3,5.4,5.5, 5.6, 5.7\n"
-            + "x\t6.1, 6.2 6.3 6.4 6.5 6.6 6.7\n"
-            + "-, \t7.1\t7.2 7.3, 7.4, 7.5\t,7.6,7.7\n";
+    String data = "ScoreMatrix MyTest (example)\n" + "A\tT\tU\tt\tx\t-\n"
+            + "A,1.1,1.2,1.3,1.4, 1.5, 1.6\n"
+            + "T,2.1 2.2 2.3 2.4 2.5 2.6\n"
+            + "U\t3.1\t3.2\t3.3\t3.4\t3.5\t3.6\t\n"
+            + "t, 5.1,5.3,5.3,5.4,5.5, 5.6\n"
+            + "x\t6.1, 6.2 6.3 6.4 6.5 6.6\n"
+            + "-, \t7.1\t7.2 7.3, 7.4, 7.5\t,7.6\n";
     FileParse fp = new FileParse(data, DataSourceType.PASTE);
     ScoreMatrixFile parser = new ScoreMatrixFile(fp);
     ScoreMatrix sm = parser.parseMatrix();
 
     assertNotNull(sm);
-    assertEquals(sm.getName(), "MyTest");
+    assertEquals(sm.getName(), "MyTest (example)");
+    assertEquals(sm.getSize(), 6);
+    assertNull(sm.getDescription());
     assertTrue(sm.isDNA());
     assertFalse(sm.isProtein());
+    assertEquals(sm.getMinimumScore(), 1.1f);
     assertEquals(sm.getPairwiseScore('A', 'A'), 1.1f);
     assertEquals(sm.getPairwiseScore('A', 'T'), 1.2f);
     assertEquals(sm.getPairwiseScore('a', 'T'), 1.2f); // A/a equivalent
-    assertEquals(sm.getPairwiseScore('A', 't'), 1.5f); // T/t not equivalent
-    assertEquals(sm.getPairwiseScore('a', 't'), 1.5f);
-    assertEquals(sm.getPairwiseScore('T', ' '), 2.4f);
-    assertEquals(sm.getPairwiseScore('U', 'x'), 3.6f);
-    assertEquals(sm.getPairwiseScore('u', 'x'), 3.6f);
-    assertEquals(sm.getPairwiseScore('U', 'X'), 0f); // X (upper) unmapped
-    assertEquals(sm.getPairwiseScore('A', '.'), 0f); // . unmapped
-    assertEquals(sm.getPairwiseScore('-', '-'), 7.7f);
+    assertEquals(sm.getPairwiseScore('A', 't'), 1.4f); // T/t not equivalent
+    assertEquals(sm.getPairwiseScore('a', 't'), 1.4f);
+    assertEquals(sm.getPairwiseScore('U', 'x'), 3.5f);
+    assertEquals(sm.getPairwiseScore('u', 'x'), 3.5f);
+    // X (upper) and '.' unmapped - get minimum score
+    assertEquals(sm.getPairwiseScore('U', 'X'), 1.1f);
+    assertEquals(sm.getPairwiseScore('A', '.'), 1.1f);
+    assertEquals(sm.getPairwiseScore('-', '-'), 7.6f);
     assertEquals(sm.getPairwiseScore('A', (char) 128), 0f); // out of range
-
-    /*
-     * without guide symbols on score rows
-     */
-    data = "ScoreMatrix MyTest\nXY\n1 2\n3 4\n";
-    fp = new FileParse(data, DataSourceType.PASTE);
-    parser = new ScoreMatrixFile(fp);
-    sm = parser.parseMatrix();
-    assertNotNull(sm);
-    assertEquals(sm.getPairwiseScore('X', 'X'), 1f);
-    assertEquals(sm.getPairwiseScore('X', 'y'), 2f);
-    assertEquals(sm.getPairwiseScore('y', 'x'), 3f);
-    assertEquals(sm.getPairwiseScore('y', 'Y'), 4f);
-    assertEquals(sm.getPairwiseScore('D', 'R'), 0f);
   }
 
   @Test(groups = "Functional")
-  public void testParse_headerMissing()
+  public void testParseMatrix_headerMissing()
   {
     String data;
 
-    data = "XY\n1 2\n3 4\n";
+    data = "X Y\n1 2\n3 4\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -93,9 +91,9 @@ public class ScoreMatrixFileTest
   }
 
   @Test(groups = "Functional")
-  public void testParse_notEnoughRows()
+  public void testParseMatrix_ncbiNotEnoughRows()
   {
-    String data = "ScoreMatrix MyTest\nXY\n1 2\n";
+    String data = "ScoreMatrix MyTest\nX Y Z\n1 2 3\n4 5 6\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -104,14 +102,14 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Expected 2 rows of score data in score matrix but only found 1");
+              "Expected 3 rows of score data in score matrix but only found 2");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_notEnoughColumns()
+  public void testParseMatrix_ncbiNotEnoughColumns()
   {
-    String data = "ScoreMatrix MyTest\nXY\n1 2\n3\n";
+    String data = "ScoreMatrix MyTest\nX Y Z\n1 2 3\n4 5\n7 8 9\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -120,17 +118,17 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Expected 2 scores at line 4 but found 1");
+              "Expected 3 scores at line 4: '4 5' but found 2");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_tooManyColumns()
+  public void testParseMatrix_ncbiTooManyColumns()
   {
     /*
      * with two too many columns:
      */
-    String data = "ScoreMatrix MyTest\nXY\n1 2\n3 4 5 6\n";
+    String data = "ScoreMatrix MyTest\nX\tY\tZ\n1 2 3\n4 5 6 7\n8 9 10\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -139,13 +137,13 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Expected 2 scores at line 4 but found 4");
+              "Expected 3 scores at line 4: '4 5 6 7' but found 4");
     }
 
     /*
      * with guide character and one too many columns:
      */
-    data = "ScoreMatrix MyTest\nXY\nX 1 2\nY 3 4 5\n";
+    data = "ScoreMatrix MyTest\nX Y\nX 1 2\nY 3 4 5\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -154,14 +152,13 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Expected 2 scores at line 4 but found 4");
+              "Expected 2 scores at line 4: 'Y 3 4 5' but found 3");
     }
 
     /*
-     * with no guide character and one too many columns:
-     * parser guesses the first column is the guide character
+     * with no guide character and one too many columns
      */
-    data = "ScoreMatrix MyTest\nXY\n1 2\n3 4 5\n";
+    data = "ScoreMatrix MyTest\nX Y\n1 2\n3 4 5\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -170,14 +167,14 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Error parsing score matrix at line 4, expected 'Y' but found '3'");
+              "Expected 2 scores at line 4: '3 4 5' but found 3");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_tooManyRows()
+  public void testParseMatrix_ncbiTooManyRows()
   {
-    String data = "ScoreMatrix MyTest\nXY\n1 2\n3 4\n6 7";
+    String data = "ScoreMatrix MyTest\n\tX\tY\tZ\n1 2 3\n4 5 6\n7 8 9\n10 11 12\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -186,14 +183,14 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Unexpected extra input line in score model file: '6 7'");
+              "Unexpected extra input line in score model file: '10 11 12'");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_badDelimiter()
+  public void testParseMatrix_ncbiBadDelimiter()
   {
-    String data = "ScoreMatrix MyTest\nXY\n1|2\n3|4\n";
+    String data = "ScoreMatrix MyTest\n X Y Z\n1|2|3\n4|5|6\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -202,14 +199,14 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Expected 2 scores at line 3 but found 1");
+              "Invalid score value '1|2|3' at line 3 column 0");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_badFloat()
+  public void testParseMatrix_ncbiBadFloat()
   {
-    String data = "ScoreMatrix MyTest\nXY\n1 2\n3 four\n";
+    String data = "ScoreMatrix MyTest\n\tX\tY\tZ\n1 2 3\n4 five 6\n7 8 9\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -218,14 +215,14 @@ public class ScoreMatrixFileTest
     } catch (IOException e)
     {
       assertEquals(e.getMessage(),
-              "Invalid score value 'four' at line 4 column 1");
+              "Invalid score value 'five' at line 4 column 1");
     }
   }
 
   @Test(groups = "Functional")
-  public void testParse_badGuideCharacter()
+  public void testParseMatrix_ncbiBadGuideCharacter()
   {
-    String data = "ScoreMatrix MyTest\nXY\nX 1 2\ny 3 4\n";
+    String data = "ScoreMatrix MyTest\n\tX Y\nX 1 2\ny 3 4\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -236,15 +233,27 @@ public class ScoreMatrixFileTest
       assertEquals(e.getMessage(),
               "Error parsing score matrix at line 4, expected 'Y' but found 'y'");
     }
+
+    data = "ScoreMatrix MyTest\n\tX Y\nXX 1 2\nY 3 4\n";
+    try
+    {
+      new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
+              .parseMatrix();
+      fail("expected exception");
+    } catch (IOException e)
+    {
+      assertEquals(e.getMessage(),
+              "Error parsing score matrix at line 3, expected 'X' but found 'XX'");
+    }
   }
 
   @Test(groups = "Functional")
-  public void testParse_nameMissing()
+  public void testParseMatrix_ncbiNameMissing()
   {
     /*
-     * Name missing
+     * Name missing on ScoreMatrix header line
      */
-    String data = "ScoreMatrix\nXY\n1 2\n3 4\n";
+    String data = "ScoreMatrix\nX Y\n1 2\n3 4\n";
     try
     {
       new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
@@ -252,9 +261,240 @@ public class ScoreMatrixFileTest
       fail("expected exception");
     } catch (IOException e)
     {
-      assertEquals(
-              e.getMessage(),
+      assertEquals(e.getMessage(),
               "Format error: expected 'ScoreMatrix <name>', found 'ScoreMatrix' at line 1");
     }
   }
+
+  /**
+   * Test a successful parse of a (small) score matrix file
+   * 
+   * @throws IOException
+   * @throws MalformedURLException
+   */
+  @Test(groups = "Functional")
+  public void testParseMatrix_ncbiFormat()
+          throws MalformedURLException, IOException
+  {
+    // input including comment and blank lines
+    String data = "ScoreMatrix MyTest\n#comment\n\n" + "\tA\tB\tC\n"
+            + "A\t1.0\t2.0\t3.0\n" + "B\t4.0\t5.0\t6.0\n"
+            + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    ScoreMatrix sm = parser.parseMatrix();
+
+    assertNotNull(sm);
+    assertEquals(sm.getName(), "MyTest");
+    assertEquals(parser.getMatrixName(), "MyTest");
+    assertEquals(sm.getPairwiseScore('A', 'A'), 1.0f);
+    assertEquals(sm.getPairwiseScore('B', 'c'), 6.0f);
+    assertEquals(sm.getSize(), 3);
+  }
+
+  /**
+   * Test a successful parse of a (small) score matrix file
+   * 
+   * @throws IOException
+   * @throws MalformedURLException
+   */
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaIndexBlosum80()
+          throws MalformedURLException, IOException
+  {
+    FileParse fp = new FileParse("resources/scoreModel/blosum80.scm",
+            DataSourceType.FILE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    ScoreMatrix sm = parser.parseMatrix();
+
+    assertNotNull(sm);
+    assertEquals(sm.getName(), "HENS920103");
+    assertEquals(sm.getDescription(),
+            "BLOSUM80 substitution matrix (Henikoff-Henikoff, 1992)");
+    assertFalse(sm.isDNA());
+    assertTrue(sm.isProtein());
+    assertEquals(20, sm.getSize());
+
+    assertEquals(sm.getPairwiseScore('A', 'A'), 7f);
+    assertEquals(sm.getPairwiseScore('A', 'R'), -3f);
+    assertEquals(sm.getPairwiseScore('r', 'a'), -3f); // A/a equivalent
+  }
+
+  /**
+   * Test a successful parse of a (small) score matrix file
+   * 
+   * @throws IOException
+   * @throws MalformedURLException
+   */
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindexFormat()
+          throws MalformedURLException, IOException
+  {
+    /*
+     * aaindex format has scores for diagonal and below only
+     */
+    String data = "H MyTest\n" + "D My description\n" + "R PMID:1438297\n"
+            + "A Authors, names\n" + "T Journal title\n"
+            + "J Journal reference\n" + "* matrix in 1/3 Bit Units\n"
+            + "M rows = ABC, cols = ABC\n" + "A\t1.0\n" + "B\t4.0\t5.0\n"
+            + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    ScoreMatrix sm = parser.parseMatrix();
+
+    assertNotNull(sm);
+    assertEquals(sm.getSize(), 3);
+    assertEquals(sm.getName(), "MyTest");
+    assertEquals(sm.getDescription(), "My description");
+    assertEquals(sm.getPairwiseScore('A', 'A'), 1.0f);
+    assertEquals(sm.getPairwiseScore('A', 'B'), 4.0f);
+    assertEquals(sm.getPairwiseScore('A', 'C'), 7.0f);
+    assertEquals(sm.getPairwiseScore('B', 'A'), 4.0f);
+    assertEquals(sm.getPairwiseScore('B', 'B'), 5.0f);
+    assertEquals(sm.getPairwiseScore('B', 'C'), 8.0f);
+    assertEquals(sm.getPairwiseScore('C', 'C'), 9.0f);
+    assertEquals(sm.getPairwiseScore('C', 'B'), 8.0f);
+    assertEquals(sm.getPairwiseScore('C', 'A'), 7.0f);
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindex_mMissing()
+          throws MalformedURLException, IOException
+  {
+    /*
+     * aaindex format but M cols=, rows= is missing
+     */
+    String data = "H MyTest\n" + "A\t1.0\n" + "B\t4.0\t5.0\n"
+            + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    try
+    {
+      parser.parseMatrix();
+      fail("Expected exception");
+    } catch (FileFormatException e)
+    {
+      assertEquals(e.getMessage(), "No alphabet specified in matrix file");
+    }
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindex_rowColMismatch()
+          throws MalformedURLException, IOException
+  {
+    String data = "H MyTest\n" + "M rows=ABC, cols=ABD\n" + "A\t1.0\n"
+            + "B\t4.0\t5.0\n" + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    try
+    {
+      parser.parseMatrix();
+      fail("Expected exception");
+    } catch (FileFormatException e)
+    {
+      assertEquals(e.getMessage(),
+              "Unexpected aaIndex score matrix data at line 2: M rows=ABC, cols=ABD rows != cols");
+    }
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_ncbiHeaderRepeated()
+  {
+    String data = "ScoreMatrix BLOSUM\nScoreMatrix PAM250\nX Y\n1 2\n3 4\n";
+    try
+    {
+      new ScoreMatrixFile(new FileParse(data, DataSourceType.PASTE))
+              .parseMatrix();
+      fail("expected exception");
+    } catch (IOException e)
+    {
+      assertEquals(e.getMessage(),
+              "Error: 'ScoreMatrix' repeated in file at line 2");
+    }
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindex_tooManyRows()
+          throws MalformedURLException, IOException
+  {
+    String data = "H MyTest\n" + "M rows=ABC, cols=ABC\n" + "A\t1.0\n"
+            + "B\t4.0\t5.0\n" + "C\t7.0\t8.0\t9.0\n" + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    try
+    {
+      parser.parseMatrix();
+      fail("Expected exception");
+    } catch (FileFormatException e)
+    {
+      assertEquals(e.getMessage(), "Too many data rows in matrix file");
+    }
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindex_extraDataLines()
+          throws MalformedURLException, IOException
+  {
+    String data = "H MyTest\n" + "M rows=ABC, cols=ABC\n" + "A\t1.0\n"
+            + "B\t4.0\t5.0\n" + "C\t7.0\t8.0\t9.0\n" + "something extra\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    try
+    {
+      parser.parseMatrix();
+      fail("Expected exception");
+    } catch (FileFormatException e)
+    {
+      assertEquals(e.getMessage(), "Too many data rows in matrix file");
+    }
+  }
+
+  @Test(groups = "Functional")
+  public void testParseMatrix_aaindex_tooFewColumns()
+          throws MalformedURLException, IOException
+  {
+    String data = "H MyTest\n" + "M rows=ABC, cols=ABC\n" + "A\t1.0\n"
+            + "B\t4.0\t5.0\n" + "C\t7.0\t8.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+    try
+    {
+      parser.parseMatrix();
+      fail("Expected exception");
+    } catch (FileFormatException e)
+    {
+      assertEquals(e.getMessage(),
+              "Expected 3 scores at line 5: 'C\t7.0\t8.0' but found 2");
+    }
+  }
+
+  /**
+   * Test a successful parse and register of a score matrix file
+   * 
+   * @throws IOException
+   * @throws MalformedURLException
+   */
+  @Test(groups = "Functional")
+  public void testParse_ncbiFormat()
+          throws MalformedURLException, IOException
+  {
+    assertNull(ScoreModels.getInstance().getScoreModel("MyNewTest", null));
+
+    String data = "ScoreMatrix MyNewTest\n" + "\tA\tB\tC\n"
+            + "A\t1.0\t2.0\t3.0\n" + "B\t4.0\t5.0\t6.0\n"
+            + "C\t7.0\t8.0\t9.0\n";
+    FileParse fp = new FileParse(data, DataSourceType.PASTE);
+    ScoreMatrixFile parser = new ScoreMatrixFile(fp);
+
+    parser.parse();
+
+    ScoreMatrix sm = (ScoreMatrix) ScoreModels.getInstance()
+            .getScoreModel("MyNewTest", null);
+    assertNotNull(sm);
+    assertEquals(sm.getName(), "MyNewTest");
+    assertEquals(parser.getMatrixName(), "MyNewTest");
+    assertEquals(sm.getPairwiseScore('A', 'A'), 1.0f);
+    assertEquals(sm.getPairwiseScore('B', 'c'), 6.0f);
+    assertEquals(sm.getSize(), 3);
+  }
 }