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 jalview.datamodel.HMMNode;
import jalview.datamodel.HiddenMarkovModel;
import java.io.FileNotFoundException;
import java.io.FileReader;
import java.io.IOException;
-import java.io.PrintWriter;
import java.util.ArrayList;
import java.util.List;
import java.util.Scanner;
+import org.testng.annotations.BeforeClass;
import org.testng.annotations.Test;
-public class HMMFileTest {
-
+import junit.extensions.PA;
+public class HMMFileTest {
- HMMFile fn3 = new HMMFile(
- new FileParse("test/jalview/io/test_fn3_hmm.txt",
- DataSourceType.FILE));
+ HMMFile fn3;
- HMMFile pKinase = new HMMFile(
- new FileParse("test/jalview/io/test_PKinase_hmm.txt",
- DataSourceType.FILE));
+ HMMFile pKinase;
- HMMFile made1 = new HMMFile(
- new FileParse("test/jalview/io/test_MADE1_hmm.txt",
- DataSourceType.FILE));
+ HMMFile made1;
- HMMFileTest() throws IOException
+ @BeforeClass(alwaysRun = true)
+ public void setUp() throws IOException
{
+ fn3 = new HMMFile("test/jalview/io/test_fn3_hmm.txt",
+ DataSourceType.FILE);
- }
-
-
+ pKinase = new HMMFile("test/jalview/io/test_PKinase_hmm.txt",
+ DataSourceType.FILE);
+ made1 = new HMMFile("test/jalview/io/test_MADE1_hmm.txt",
+ DataSourceType.FILE);
+ }
- @Test
+ @Test(groups = "Functional")
public void testParse() throws IOException
{
-
- pKinase.parse();
HiddenMarkovModel hmm = pKinase.getHMM();
assertEquals(hmm.getName(), "Pkinase");
- assertEquals(hmm.getAccessionNumber(), "PF00069.17");
- assertEquals(hmm.getDescription(), "Protein kinase domain");
- assertEquals(hmm.getLength().intValue(), 260);
- assertNull(hmm.getMaxInstanceLength());
+ assertEquals(hmm.getProperty(HMMFile.ACCESSION_NUMBER), "PF00069.17");
+ assertEquals(hmm.getProperty(HMMFile.DESCRIPTION),
+ "Protein kinase domain");
+ assertEquals(hmm.getLength(), 260);
+ assertNull(hmm.getProperty(HMMFile.MAX_LENGTH));
assertEquals(hmm.getAlphabetType(), "amino");
- assertEquals(hmm.referenceAnnotationIsActive(), false);
- assertEquals(hmm.maskValueIsActive(), false);
- assertEquals(hmm.consensusResidueIsActive(), true);
- assertEquals(hmm.consensusStructureIsActive(),
- true);
- assertEquals(hmm.mapIsActive(), true);
- assertEquals(hmm.getDate(), "Thu Jun 16 11:44:06 2011");
- assertNull(hmm.getCommandLineLog());
- assertEquals(hmm.getNumberOfSequences().intValue(), 54);
- assertEquals(hmm.getEffectiveNumberOfSequences(), 3.358521, 4d);
- assertEquals(hmm.getCheckSum().longValue(), 3106786190l);
- assertEquals(hmm.getGatheringThreshold(), "70.30 70.30");
- assertEquals(hmm.getTrustedCutoff(), "70.30 70.30");
- assertEquals(hmm.getNoiseCutoff(), "70.20 70.20");
-
- List<Character> symbols = new ArrayList<>();
- symbols.add('A');
- symbols.add('C');
- symbols.add('D');
- symbols.add('E');
- symbols.add('F');
- symbols.add('G');
- symbols.add('H');
- symbols.add('I');
- symbols.add('K');
- symbols.add('L');
- symbols.add('M');
- symbols.add('N');
- symbols.add('P');
- symbols.add('Q');
- symbols.add('R');
- symbols.add('S');
- symbols.add('T');
- symbols.add('V');
- symbols.add('W');
- symbols.add('Y');
-
- assertEquals(hmm.getSymbols(), symbols);
-
- assertEquals(getMatchEmission(0, 19, hmm), 0.032298, 0.001d);
- assertEquals(getMatchEmission(12, 12, hmm), 0.0130, 0.001d);
- assertEquals(getMatchEmission(23, 7, hmm), 0.02583, 0.001d);
- assertEquals(getMatchEmission(54, 1, hmm), 0.008549, 0.001d);
- assertEquals(getMatchEmission(178, 3, hmm), 0.07998, 0.001d);
- assertEquals(getMatchEmission(210, 2, hmm), 0.014465, 0.001d);
- assertEquals(getMatchEmission(260, 19, hmm), 0.02213, 0.001d);
-
- assertEquals(getInsertEmission(2, 1, hmm), 0.012, 0.001d);
- assertEquals(getInsertEmission(15, 6, hmm), 0.02411, 0.001d);
- assertEquals(getInsertEmission(22, 9, hmm), 0.06764, 0.001d);
- assertEquals(getInsertEmission(57, 2, hmm), 0.0623, 0.001d);
- assertEquals(getInsertEmission(203, 16, hmm), 0.0623, 0.001d);
- assertEquals(getInsertEmission(255, 12, hmm), 0.0647, 0.001d);
-
- assertEquals(getStateTransition(0, 6, hmm),
- Double.NEGATIVE_INFINITY);
- assertEquals(getStateTransition(3, 6, hmm), 0.3848, 0.001d);
- assertEquals(getStateTransition(29, 3, hmm), 0.5382, 0.001d);
- assertEquals(getStateTransition(169, 3, hmm), 0.2916, 0.001d);
- assertEquals(getStateTransition(209, 0, hmm), 0.99, 0.001d);
- assertEquals(getStateTransition(243, 1, hmm), 0.0066, 0.001d);
-
- assertEquals(hmm.getNodeAlignmentColumn(3).intValue(), 2);
+ assertFalse(hmm.getBooleanProperty(HMMFile.REFERENCE_ANNOTATION));
+ assertFalse(hmm.getBooleanProperty(HMMFile.MASKED_VALUE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.CONSENSUS_RESIDUE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.CONSENSUS_STRUCTURE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.MAP));
+ assertEquals(hmm.getProperty(HMMFile.DATE), "Thu Jun 16 11:44:06 2011");
+ assertNull(hmm.getProperty(HMMFile.COMMAND_LOG));
+ assertEquals(hmm.getProperty(HMMFile.NUMBER_OF_SEQUENCES), "54");
+ assertEquals(hmm.getProperty(HMMFile.EFF_NUMBER_OF_SEQUENCES),
+ "3.358521");
+ assertEquals(hmm.getProperty(HMMFile.CHECK_SUM), "3106786190");
+ assertEquals(hmm.getProperty(HMMFile.GATHERING_THRESHOLD),
+ "70.30 70.30");
+ assertEquals(hmm.getProperty(HMMFile.TRUSTED_CUTOFF), "70.30 70.30");
+ assertEquals(hmm.getProperty(HMMFile.NOISE_CUTOFF), "70.20 70.20");
+
+ assertEquals(hmm.getSymbols(), "ACDEFGHIKLMNPQRSTVWY");
+
+ assertEquals(hmm.getMatchEmissionProbability(0, 'Y'), 0.16102, 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(11, 'P'), 0.0130, 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(24, 'I'), 0.02583, 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(83, 'C'), 0.008549,
+ 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(332, 'E'), 0.07998,
+ 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(381, 'D'), 0.014465,
+ 0.001d);
+ assertEquals(hmm.getMatchEmissionProbability(475, 'Y'), 0.02213,
+ 0.001d);
+
+ assertEquals(hmm.getInsertEmissionProbability(1, 'C'), 0.012, 0.001d);
+ assertEquals(hmm.getInsertEmissionProbability(14, 'H'), 0.02411,
+ 0.001d);
+ assertEquals(hmm.getInsertEmissionProbability(23, 'L'), 0.06764,
+ 0.001d);
+ assertEquals(hmm.getInsertEmissionProbability(90, 'D'), 0.0623, 0.001d);
+ assertEquals(hmm.getInsertEmissionProbability(374, 'T'), 0.0623,
+ 0.001d);
+ assertEquals(hmm.getInsertEmissionProbability(470, 'P'), 0.0647,
+ 0.001d);
+
+ assertEquals(hmm.getStateTransitionProbability(2, 6), 0.3848, 0.001d);
+ assertEquals(hmm.getStateTransitionProbability(38, 3), 0.5382, 0.001d);
+ assertEquals(hmm.getStateTransitionProbability(305, 3), 0.2916, 0.001d);
+ assertEquals(hmm.getStateTransitionProbability(380, 0), 0.99, 0.001d);
+ assertEquals(hmm.getStateTransitionProbability(453, 1), 0.0066, 0.001d);
+
+ assertEquals(hmm.getNodeMapPosition(3), 3);
assertEquals(hmm.getReferenceAnnotation(7), '-');
assertEquals(hmm.getConsensusResidue(23), 't');
assertEquals(hmm.getMaskedValue(30), '-');
assertEquals(hmm.getConsensusStructure(56), 'S');
- assertEquals(hmm.getNodeAlignmentColumn(78).intValue(), 135);
+ assertEquals(hmm.getNodeMapPosition(78), 136);
assertEquals(hmm.getReferenceAnnotation(93), '-');
assertEquals(hmm.getConsensusResidue(145), 'a');
assertEquals(hmm.getMaskedValue(183), '-');
assertEquals(hmm.getConsensusStructure(240), 'H');
-
+ }
+
+ /**
+ * Test that Jalview can parse an HMM file even with a bunch of 'mandatory'
+ * fields missing (including no MAP annotation or // terminator line)
+ *
+ * @throws IOException
+ */
+ @Test(groups = "Functional")
+ public void testParse_minimalFile() throws IOException
+ {
+ /*
+ * ALPH is absent, alphabet inferred from HMM header line
+ * Optional COMPO line is absent
+ * first line after HMM is a guide line for readability
+ * next line is BEGIN node insert emissions
+ * next line is BEGIN node transitions
+ * next line is first sequence node match emissions 1.1 1.2 1.3
+ * next line is first sequence node insert emissions 1.4 1.5 1.6
+ * last line is first sequence node transitions
+ */
+ //@formatter:off
+ String hmmData =
+ "HMMER3\n" +
+ "HMM P M J\n" +
+ // both spec and parser require a line after the HMM line
+ " m->m m->i m->d i->m i->i d->m d->d\n" +
+ " 0.1 0.2 0.3\n" +
+ " 0.4 0.5 0.6 0.7 0.8 0.9 0.95\n" +
+ " 1 1.1 1.2 1.3 - - - - -\n" +
+ " 1.4 1.5 1.6\n" +
+ " 1.7 1.8 1.9 2.0 2.1 2.2 2.3\n" +
+ " 2 1.01 1.02 1.03 - - - - -\n" +
+ " 1.04 1.05 1.06\n" +
+ " 1.7 1.8 1.9 2.0 2.1 2.2 2.3\n";
+ //@formatter:on
+ HMMFile parser = new HMMFile(hmmData, DataSourceType.PASTE);
+ HiddenMarkovModel hmm = parser.getHMM();
+ assertNotNull(hmm);
+ assertEquals(hmm.getSymbols(), "PMJ");
+ // no LENG property: this should return node count excluding BEGIN node
+ assertEquals(hmm.getLength(), 2);
+
+ // node 1 (implicitly mapped to column 0)
+ double prob = hmm.getMatchEmissionProbability(0, 'p');
+ assertEquals(prob, Math.pow(Math.E, -1.1));
+ prob = hmm.getInsertEmissionProbability(0, 'J');
+ assertEquals(prob, Math.pow(Math.E, -1.6));
+
+ // node 2 (implicitly mapped to column 1)
+ prob = hmm.getMatchEmissionProbability(1, 'M');
+ assertEquals(prob, Math.pow(Math.E, -1.02));
+ prob = hmm.getInsertEmissionProbability(1, 'm');
+ assertEquals(prob, Math.pow(Math.E, -1.05));
}
- @Test
- public void testParseFileProperties() throws IOException
+ @Test(groups = "Functional")
+ public void testParseHeaderLines_amino() throws IOException
{
FileReader fr = new FileReader(
new File("test/jalview/io/test_fn3_hmm.txt"));
BufferedReader br = new BufferedReader(fr);
- fn3.parseFileProperties(br);
- fn3.parseModel(br); // this is for a later test
- HiddenMarkovModel testHMM = new HiddenMarkovModel();
- testHMM = fn3.getHMM();
+ HiddenMarkovModel hmm = new HiddenMarkovModel();
+ HMMFile testee = new HMMFile();
+ PA.setValue(testee, "hmm", hmm);
+ testee.parseHeaderLines(br);
br.close();
fr.close();
- assertEquals(testHMM.getName(), "fn3");
- assertEquals(testHMM.getAccessionNumber(), "PF00041.13");
- assertEquals(testHMM.getDescription(),
+ assertEquals(hmm.getName(), "fn3");
+ assertEquals(hmm.getProperty(HMMFile.ACCESSION_NUMBER), "PF00041.13");
+ assertEquals(hmm.getProperty(HMMFile.DESCRIPTION),
"Fibronectin type III domain");
- assertEquals(testHMM.getLength().intValue(), 86);
- assertNull(testHMM.getMaxInstanceLength());
- assertEquals(testHMM.getAlphabetType(), "amino");
- assertEquals(testHMM.referenceAnnotationIsActive(), false);
- assertEquals(testHMM.maskValueIsActive(), false);
- assertEquals(testHMM.consensusResidueIsActive(), true);
- assertEquals(testHMM.consensusStructureIsActive(), true);
- assertEquals(testHMM.mapIsActive(), true);
- assertEquals(testHMM.getDate(), "Fri Jun 20 08:22:31 2014");
- assertNull(testHMM.getCommandLineLog());
- assertEquals(testHMM.getNumberOfSequences().intValue(), 106);
- assertEquals(testHMM.getEffectiveNumberOfSequences(), 11.415833, 4d);
- assertEquals(testHMM.getCheckSum().longValue(), 3564431818l);
- assertEquals(testHMM.getGatheringThreshold(), "8.00 7.20");
- assertEquals(testHMM.getTrustedCutoff(), "8.00 7.20");
- assertEquals(testHMM.getNoiseCutoff(), "7.90 7.90");
- assertEquals(testHMM.getViterbi(), "-9.7737 0.71847");
- assertEquals(testHMM.getMSV(), "-9.4043 0.71847");
- assertEquals(testHMM.getForward(), "-3.8341 0.71847");
-
-
- FileReader fr3 = new FileReader(
+ assertEquals(hmm.getProperty(HMMFile.LENGTH), "86");
+ assertNull(hmm.getProperty(HMMFile.MAX_LENGTH));
+ assertEquals(hmm.getAlphabetType(), "amino");
+ assertFalse(hmm.getBooleanProperty(HMMFile.REFERENCE_ANNOTATION));
+ assertFalse(hmm.getBooleanProperty(HMMFile.MASKED_VALUE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.CONSENSUS_RESIDUE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.CONSENSUS_STRUCTURE));
+
+ assertTrue(hmm.getBooleanProperty(HMMFile.MAP));
+ assertEquals(hmm.getProperty(HMMFile.DATE), "Fri Jun 20 08:22:31 2014");
+ assertNull(hmm.getProperty(HMMFile.COMMAND_LOG));
+ assertEquals(hmm.getProperty(HMMFile.NUMBER_OF_SEQUENCES), "106");
+ assertEquals(hmm.getProperty(HMMFile.EFF_NUMBER_OF_SEQUENCES),
+ "11.415833");
+ assertEquals(hmm.getProperty(HMMFile.CHECK_SUM), "3564431818");
+ assertEquals(hmm.getProperty(HMMFile.GATHERING_THRESHOLD), "8.00 7.20");
+ assertEquals(hmm.getProperty(HMMFile.TRUSTED_CUTOFF), "8.00 7.20");
+ assertEquals(hmm.getProperty(HMMFile.NOISE_CUTOFF), "7.90 7.90");
+ assertEquals(hmm.getViterbi(), "-9.7737 0.71847");
+ assertEquals(hmm.getMSV(), "-9.4043 0.71847");
+ assertEquals(hmm.getForward(), "-3.8341 0.71847");
+ }
+
+ @Test(groups = "Functional")
+ public void testParseHeaderLines_dna() throws IOException
+ {
+ FileReader fr = new FileReader(
new File("test/jalview/io/test_MADE1_hmm.txt"));
- BufferedReader br3 = new BufferedReader(fr3);
- made1.parseFileProperties(br3);
- testHMM = made1.getHMM();
- br3.close();
- fr3.close();
+ BufferedReader br = new BufferedReader(fr);
+ HiddenMarkovModel hmm = new HiddenMarkovModel();
+ HMMFile testee = new HMMFile();
+ PA.setValue(testee, "hmm", hmm);
+ testee.parseHeaderLines(br);
+ br.close();
+ fr.close();
- assertEquals(testHMM.getName(), "MADE1");
- assertEquals(testHMM.getAccessionNumber(), "DF0000629.2");
- assertEquals(testHMM.getDescription(),
+ assertEquals(hmm.getName(), "MADE1");
+ assertEquals(hmm.getProperty(HMMFile.ACCESSION_NUMBER),
+ "DF0000629.2");
+ assertEquals(hmm.getProperty(HMMFile.DESCRIPTION),
"MADE1 (MAriner Derived Element 1), a TcMar-Mariner DNA transposon");
- assertEquals(testHMM.getLength().intValue(), 80);
- assertEquals(testHMM.getMaxInstanceLength().intValue(), 426);
- assertEquals(testHMM.getAlphabetType(), "DNA");
- assertEquals(testHMM.referenceAnnotationIsActive(), true);
- assertEquals(testHMM.maskValueIsActive(), false);
- assertEquals(testHMM.consensusResidueIsActive(), true);
- assertEquals(testHMM.consensusStructureIsActive(), false);
- assertEquals(testHMM.mapIsActive(), true);
- assertEquals(testHMM.getDate(), "Tue Feb 19 20:33:41 2013");
- assertNull(testHMM.getCommandLineLog());
- assertEquals(testHMM.getNumberOfSequences().intValue(), 1997);
- assertEquals(testHMM.getEffectiveNumberOfSequences(), 3.911818, 4d);
- assertEquals(testHMM.getCheckSum().longValue(), 3015610723l);
- assertEquals(testHMM.getGatheringThreshold(), "2.324 4.234");
- assertEquals(testHMM.getTrustedCutoff(), "2.343 1.212");
- assertEquals(testHMM.getNoiseCutoff(), "2.354 5.456");
- assertEquals(testHMM.getViterbi(), "-9.3632 0.71858");
- assertEquals(testHMM.getMSV(), "-8.5786 0.71858");
- assertEquals(testHMM.getForward(), "-3.4823 0.71858");
-
-
+ assertEquals(hmm.getProperty(HMMFile.LENGTH), "80");
+ assertEquals(hmm.getProperty(HMMFile.MAX_LENGTH), "426");
+ assertEquals(hmm.getAlphabetType(), "DNA");
+ assertTrue(hmm.getBooleanProperty(HMMFile.REFERENCE_ANNOTATION));
+ assertFalse(hmm.getBooleanProperty(HMMFile.MASKED_VALUE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.CONSENSUS_RESIDUE));
+ assertFalse(hmm.getBooleanProperty(HMMFile.CONSENSUS_STRUCTURE));
+ assertTrue(hmm.getBooleanProperty(HMMFile.MAP));
+ assertEquals(hmm.getProperty(HMMFile.DATE), "Tue Feb 19 20:33:41 2013");
+ assertNull(hmm.getProperty(HMMFile.COMMAND_LOG));
+ assertEquals(hmm.getProperty(HMMFile.NUMBER_OF_SEQUENCES), "1997");
+ assertEquals(hmm.getProperty(HMMFile.EFF_NUMBER_OF_SEQUENCES), "3.911818");
+ assertEquals(hmm.getProperty(HMMFile.CHECK_SUM), "3015610723");
+ assertEquals(hmm.getProperty(HMMFile.GATHERING_THRESHOLD),
+ "2.324 4.234");
+ assertEquals(hmm.getProperty(HMMFile.TRUSTED_CUTOFF), "2.343 1.212");
+ assertEquals(hmm.getProperty(HMMFile.NOISE_CUTOFF), "2.354 5.456");
+ assertEquals(hmm.getViterbi(), "-9.3632 0.71858");
+ assertEquals(hmm.getMSV(), "-8.5786 0.71858");
+ assertEquals(hmm.getForward(), "-3.4823 0.71858");
}
- @Test
+ @Test(groups = "Functional")
public void testFillList() throws IOException
{
Scanner scanner1 = new Scanner("1.3 2.4 5.3 3.9 9.8 4.7 4.3 2.3 6.9");
filledArray.add(0.10026);
filledArray.add(0.001);
- List<Double> testList = HMMFile.fillList(scanner1, 9);
+ double[] testList = HMMFile.parseDoubles(scanner1, 9);
for (int i = 0; i < 9; i++)
{
- assertEquals(testList.get(i), filledArray.get(i), 0.001d);
-
+ assertEquals(testList[i], filledArray.get(i), 0.001d);
}
filledArray.clear();
filledArray.add(0.00355);
filledArray.add(0.2466);
- testList = HMMFile.fillList(scanner2, 5);
+ testList = HMMFile.parseDoubles(scanner2, 5);
for (int i = 0; i < 5; i++)
{
- assertEquals(testList.get(i), filledArray.get(i), 0.001d);
+ assertEquals(testList[i], filledArray.get(i), 0.001d);
}
-
}
- @Test
+ @Test(groups = "Functional")
public void testParseModel() throws IOException
{
FileReader fr = new FileReader(
new File("test/jalview/io/test_MADE1_hmm.txt"));
BufferedReader br = new BufferedReader(fr);
HiddenMarkovModel testHMM = new HiddenMarkovModel();
- for (int i = 0; i < 24; i++)
+ String line = null;
+ do
{
- br.readLine();
- }
+ line = br.readLine(); // skip header lines up to HMM plus one
+ } while (!line.startsWith("HMM "));
+ br.readLine();
+
made1.parseModel(br);
testHMM = made1.getHMM();
+
br.close();
fr.close();
- assertEquals(getMatchEmission(0, 2, testHMM), 0.1961, 0.001d);
- assertEquals(getMatchEmission(2, 1, testHMM), 0.09267, 0.001d);
- assertEquals(getMatchEmission(12, 2, testHMM), 0.07327, 0.001d);
- assertEquals(getMatchEmission(69, 1, testHMM), 0.04184, 0.001d);
- assertEquals(getMatchEmission(76, 2, testHMM), 0.07, 0.001d);
-
- assertEquals(getInsertEmission(0, 1, testHMM), 0.25, 0.001d);
- assertEquals(getInsertEmission(1, 2, testHMM), 0.25, 0.001d);
- assertEquals(getInsertEmission(31, 3, testHMM), 0.2776, 0.001d);
- assertEquals(getInsertEmission(70, 3, testHMM), 0.25, 0.001d);
- assertEquals(getInsertEmission(80, 3, testHMM), 0.25, 0.001d);
+ assertEquals(testHMM.getMatchEmissionProbability(1, 'C'), 0.09267,
+ 0.001d);
+ assertEquals(testHMM.getMatchEmissionProbability(25, 'G'), 0.07327,
+ 0.001d);
+ assertEquals(testHMM.getMatchEmissionProbability(1092, 'C'), 0.04184,
+ 0.001d);
+ assertEquals(testHMM.getMatchEmissionProbability(1107, 'G'), 0.07,
+ 0.001d);
+
+ assertEquals(testHMM.getInsertEmissionProbability(0, 'G'), 0.25,
+ 0.001d);
+ assertEquals(testHMM.getInsertEmissionProbability(247, 'T'), 0.2776,
+ 0.001d);
+ assertEquals(testHMM.getInsertEmissionProbability(1096, 'T'), 0.25,
+ 0.001d);
+ assertEquals(testHMM.getInsertEmissionProbability(1111, 'T'), 0.25,
+ 0.001d);
- assertEquals(getStateTransition(2, 0, testHMM), 0.9634, 0.001d);
- assertEquals(getStateTransition(6, 1, testHMM), 0.0203, 0.001d);
- assertEquals(getStateTransition(9, 3, testHMM), 0.2515, 0.001d);
- assertEquals(getStateTransition(20, 4, testHMM), 0.78808, 0.001d);
- assertEquals(getStateTransition(68, 2, testHMM), 0.01845, 0.001d);
- assertEquals(getStateTransition(80, 6, testHMM),
+ assertEquals(testHMM.getStateTransitionProbability(1, 0), 0.9634,
+ 0.001d);
+ assertEquals(testHMM.getStateTransitionProbability(5, 1), 0.0203,
+ 0.001d);
+ assertEquals(testHMM.getStateTransitionProbability(14, 3), 0.2515,
+ 0.001d);
+ assertEquals(testHMM.getStateTransitionProbability(65, 4), 0.78808,
+ 0.001d);
+ assertEquals(testHMM.getStateTransitionProbability(1080, 2), 0.01845,
+ 0.001d);
+ assertEquals(testHMM.getStateTransitionProbability(1111, 6),
Double.NEGATIVE_INFINITY);
-
}
- @Test
+ @Test(groups = "Functional")
public void testParseAnnotations()
{
HMMFile testFile = new HMMFile();
- testFile.getHMM().getNodes().add(new HMMNode());
- testFile.getHMM().getNodes().add(new HMMNode());
- testFile.getHMM().getNodes().add(new HMMNode());
-
-
- testFile.getHMM().setConsensusResidueStatus(true);
- testFile.getHMM().setMAPStatus(true);
- testFile.getHMM().setReferenceAnnotationStatus(true);
- testFile.getHMM().setConsensusStructureStatus(true);
- testFile.getHMM().setMaskedValueStatus(true);
+ HiddenMarkovModel hmm = new HiddenMarkovModel();
+ PA.setValue(testFile, "hmm", hmm);
+
+ List<HMMNode> nodes = new ArrayList<>();
+ nodes.add(new HMMNode()); // BEGIN node
+
+ hmm.setProperty(HMMFile.CONSENSUS_RESIDUE, "yes");
+ hmm.setProperty(HMMFile.MAP, "yes");
+ hmm.setProperty(HMMFile.REFERENCE_ANNOTATION, "yes");
+ hmm.setProperty(HMMFile.CONSENSUS_STRUCTURE, "yes");
+ hmm.setProperty(HMMFile.MASKED_VALUE, "yes");
Scanner scanner = new Scanner("1345 t t t t");
- testFile.parseAnnotations(scanner, 1);
-
- testFile.getHMM().setConsensusResidueStatus(true);
- testFile.getHMM().setMAPStatus(false);
- testFile.getHMM().setReferenceAnnotationStatus(true);
- testFile.getHMM().setConsensusStructureStatus(false);
- testFile.getHMM().setMaskedValueStatus(false);
+ HMMNode node = new HMMNode();
+ nodes.add(node);
+ testFile.parseAnnotations(scanner, node);
+
+ hmm.setProperty(HMMFile.CONSENSUS_RESIDUE, "yes");
+ hmm.setProperty(HMMFile.MAP, "no");
+ hmm.setProperty(HMMFile.REFERENCE_ANNOTATION, "yes");
+ hmm.setProperty(HMMFile.CONSENSUS_STRUCTURE, "no");
+ hmm.setProperty(HMMFile.MASKED_VALUE, "no");
Scanner scanner2 = new Scanner("- y x - -");
- testFile.parseAnnotations(scanner2, 2);
-
- HiddenMarkovModel hmm = testFile.getHMM();
-
- assertEquals(hmm.getNodeAlignmentColumn(1).intValue(), 1344);
+ node = new HMMNode();
+ nodes.add(node);
+ testFile.parseAnnotations(scanner2, node);
+
+ hmm.setNodes(nodes);
+
+ assertEquals(hmm.getNodeMapPosition(1), 1345);
assertEquals(hmm.getConsensusResidue(1), 't');
assertEquals(hmm.getReferenceAnnotation(1), 't');
assertEquals(hmm.getMaskedValue(1), 't');
assertEquals(hmm.getConsensusStructure(1), 't');
- assertEquals(hmm.findNodeIndex(1344).intValue(), 1);
-
scanner.close();
-
}
/**
*
* @throws IOException
*/
-
-
- @Test(priority = 3)
- public void testExportFile() throws IOException
+ @Test(groups = "Functional")
+ public void testPrint_roundTrip() throws IOException
{
- pKinase.exportFile("test/jalview/io/test_export_hmm.txt");
+ String output = pKinase.print();
HMMFile pKinaseClone = new HMMFile(
- new FileParse("test/jalview/io/test_export_hmm.txt",
- DataSourceType.FILE));
- pKinaseClone.parse();
- HiddenMarkovModel pKinaseHMM = new HiddenMarkovModel();
- HiddenMarkovModel pKinaseCloneHMM = new HiddenMarkovModel();
- pKinaseHMM = pKinase.getHMM();
- pKinaseCloneHMM = pKinaseClone.getHMM();
-
- for (int i = 0; i < pKinaseHMM.getLength(); i++)
- {
- List<Double> list1;
- List<Double> list2;
- boolean result;
-
- list1 = pKinaseHMM.getNode(i).getMatchEmissions();
- list2 = pKinaseCloneHMM.getNode(i).getMatchEmissions();
-
- result = checkIfListsAreIdentical(list1, list2);
- assertEquals(result, true);
-
- list1 = pKinaseHMM.getNode(i).getInsertEmissions();
- list2 = pKinaseCloneHMM.getNode(i).getInsertEmissions();
-
- result = checkIfListsAreIdentical(list1, list2);
- assertEquals(result, true);
-
- list1 = pKinaseHMM.getNode(i).getStateTransitions();
- list2 = pKinaseCloneHMM.getNode(i).getStateTransitions();
+ new FileParse(output, DataSourceType.PASTE));
+ HiddenMarkovModel pKinaseHMM = pKinase.getHMM();
+ HiddenMarkovModel pKinaseCloneHMM = pKinaseClone.getHMM();
- result = checkIfListsAreIdentical(list1, list2);
- assertEquals(result, true);
+ checkModelsMatch(pKinaseHMM, pKinaseCloneHMM);
+ }
+
+ /**
+ * A helper method to check two HMM models have the same values
+ *
+ * @param model1
+ * @param model2
+ */
+ protected void checkModelsMatch(HiddenMarkovModel model1,
+ HiddenMarkovModel model2)
+ {
+ assertEquals(model1.getLength(), model2.getLength());
+
+ for (int i = 0; i < model1.getLength(); i++)
+ {
+ String msg = "For Node" + i;
+ assertEquals(model1.getNode(i).getMatchEmissions(),
+ model2.getNode(i).getMatchEmissions(), msg);
+ assertEquals(model1.getNode(i).getInsertEmissions(),
+ model2.getNode(i).getInsertEmissions(), msg);
+ assertEquals(model1.getNode(i).getStateTransitions(),
+ model2.getNode(i).getStateTransitions(), msg);
if (i > 0)
{
- int alignColumn1;
- int alignColumn2;
-
- alignColumn1 = pKinaseHMM.getNodeAlignmentColumn(i);
- alignColumn2 = pKinaseCloneHMM.getNodeAlignmentColumn(i);
-
- assertEquals(alignColumn1, alignColumn2);
-
- char annotation1;
- char annotation2;
-
- annotation1 = pKinaseHMM.getReferenceAnnotation(i);
- annotation2 = pKinaseCloneHMM.getReferenceAnnotation(i);
-
- assertEquals(annotation1, annotation2);
-
- annotation1 = pKinaseHMM.getConsensusResidue(i);
- annotation2 = pKinaseCloneHMM.getConsensusResidue(i);
-
- assertEquals(annotation1, annotation2);
+ assertEquals(model1.getNodeMapPosition(i),
+ model2.getNodeMapPosition(i), msg);
+ assertEquals(model1.getReferenceAnnotation(i),
+ model2.getReferenceAnnotation(i), msg);
+ assertEquals(model1.getConsensusResidue(i),
+ model2.getConsensusResidue(i), msg);
}
-
}
-
}
- @Test(priority = 1)
- public void testAppendFileProperties() throws FileNotFoundException
+ @Test(groups = "Functional")
+ public void testAppendProperties() throws FileNotFoundException
{
- PrintWriter writer = new PrintWriter(
- "test/jalview/io/test_export_hmm.txt");
- fn3.appendFileProperties(writer);
- writer.close();
- String content;
- File file = new File("test/jalview/io/test_export_hmm.txt");
+ StringBuilder sb = new StringBuilder();
+ fn3.appendProperties(sb);
- Scanner testScanner = new Scanner(file);
- testScanner.useDelimiter("\\Z");
+ Scanner testScanner = new Scanner(sb.toString());
String[] expected = new String[] { "HMMER3/f [3.1b1 | May 2013]",
"NAME fn3", "ACC PF00041.13",
testScanner.close();
}
- @Test(priority = 2)
- public void testAppendModel() throws FileNotFoundException
+ @Test(groups = "Functional")
+ public void testAppendModelAsString() throws FileNotFoundException
{
- PrintWriter writer = new PrintWriter(
- "test/jalview/io/test_export_hmm.txt");
- fn3.appendModel(writer);
- writer.close();
- String string;
- File file = new File("test/jalview/io/test_export_hmm.txt");
- Scanner scanner = new Scanner(file);
- scanner.useDelimiter("\\Z");
- string = scanner.next();
+ StringBuilder sb = new StringBuilder();
+ fn3.appendModelAsString(sb);
+ String string = sb.toString();
assertEquals(findValue(2, 2, 2, string), "4.42225");
assertEquals(findValue(12, 14, 1, string), "2.79307");
assertEquals(findValue(16, 65, 1, string), "2.81003");
assertEquals(findValue(14, 3, 1, string), "2.69012");
assertEquals(findValue(11, 32, 1, string), "4.34805");
-
}
/**
+ * A helper method to find a token in the model string
*
* @param symbolIndex
* index of symbol being searched. First symbol has index 1.
* string model being searched
* @return value at specified position
*/
-
- public String findValue(int symbolIndex, int nodeIndex, int line,
+ private String findValue(int symbolIndex, int nodeIndex, int line,
String model)
{
-
String value = "";
Scanner scanner = new Scanner(model);
scanner.nextLine();
{
scanner.next();
}
-
}
- return value;
-
- }
-
- public boolean checkIfListsAreIdentical(List<Double> list1,
- List<Double> list2)
- {
- boolean isDifferent = false;
- for (int i = 0; i < list1.size(); i++)
- {
- Double entry1;
- Double entry2;
- entry1 = list1.get(i);
- entry2 = list2.get(i);
- if (!(entry1 == entry2))
- {
- isDifferent = true;
- }
- }
- return isDifferent;
- }
-
- /**
- * gets the match emission at a node for a symbol
- *
- * @param nodeIndex
- * position of node in model
- * @param symbolIndex
- * index of symbol being searched
- * @return negative log probability of a match emission of the given symbol
- */
- public double getMatchEmission(int nodeIndex, int symbolIndex,
- HiddenMarkovModel hmm)
- {
- double value = hmm.getNodes().get(nodeIndex).getMatchEmissions()
- .get(symbolIndex);
- return value;
- }
-
- /**
- * gets the insert emission at a node for a symbol
- *
- * @param nodeIndex
- * position of node in model
- * @param symbolIndex
- * index of symbol being searched
- * @return negative log probability of an insert emission of the given symbol
- */
- public double getInsertEmission(int nodeIndex, int symbolIndex,
- HiddenMarkovModel hmm)
- {
- double value = hmm.getNodes().get(nodeIndex).getInsertEmissions()
- .get(symbolIndex);
- return value;
- }
-
- /**
- * gets the state transition at a node for a specific transition
- *
- * @param nodeIndex
- * position of node in model
- * @param transitionIndex
- * index of stransition being searched
- * @return negative log probability of a state transition of the given type
- */
- public double getStateTransition(int nodeIndex, int transitionIndex,
- HiddenMarkovModel hmm)
- {
- double value = hmm.getNodes().get(nodeIndex).getStateTransitions()
- .get(transitionIndex);
+ scanner.close();
return value;
}
-
}
git://source.jalview.org / jalview.git / blobdiff