X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=test%2Fjalview%2Fanalysis%2FAAFrequencyTest.java;h=14cc526459c6de174f53b622ed74e3cd4db5bf47;hb=c794c5033adeee182b03a5ea92c0a7495a29661f;hp=0ddbddc55cfd2d638fbee46a0ccee61a683fab38;hpb=d94db336381e48d9155736ca4ea058d9f747439c;p=jalview.git diff --git a/test/jalview/analysis/AAFrequencyTest.java b/test/jalview/analysis/AAFrequencyTest.java index 0ddbddc..14cc526 100644 --- a/test/jalview/analysis/AAFrequencyTest.java +++ b/test/jalview/analysis/AAFrequencyTest.java @@ -25,13 +25,49 @@ import static org.testng.AssertJUnit.assertNull; import jalview.datamodel.AlignmentAnnotation; import jalview.datamodel.Annotation; +import jalview.datamodel.HiddenMarkovModel; +import jalview.datamodel.Profile; +import jalview.datamodel.ProfileI; +import jalview.datamodel.Profiles; +import jalview.datamodel.ProfilesI; +import jalview.datamodel.ResidueCount; import jalview.datamodel.Sequence; import jalview.datamodel.SequenceI; +import jalview.gui.JvOptionPane; +import jalview.io.DataSourceType; +import jalview.io.FileParse; +import jalview.io.HMMFile; +import java.io.IOException; +import java.net.MalformedURLException; + +import java.util.Hashtable; + +import org.testng.annotations.BeforeClass; import org.testng.annotations.Test; public class AAFrequencyTest { + HiddenMarkovModel hmm; + + @BeforeClass(alwaysRun = true) + public void setUpJvOptionPane() + { + JvOptionPane.setInteractiveMode(false); + JvOptionPane.setMockResponse(JvOptionPane.CANCEL_OPTION); + } + + @BeforeClass(alwaysRun = true) + public void setUp() throws IOException, MalformedURLException + { + /* + * load a dna (ACGT) HMM file to a HiddenMarkovModel + */ + HMMFile hmmFile = new HMMFile(new FileParse( + "test/jalview/io/test_MADE1_hmm.txt", DataSourceType.FILE)); + hmm = hmmFile.getHMM(); + } + @Test(groups = { "Functional" }) public void testCalculate_noProfile() { @@ -40,12 +76,12 @@ public class AAFrequencyTest SequenceI seq3 = new Sequence("Seq3", "C---G"); SequenceI seq4 = new Sequence("Seq4", "CA--t"); SequenceI[] seqs = new SequenceI[] { seq1, seq2, seq3, seq4 }; - Profile[] result = new Profile[seq1.getLength()]; - - AAFrequency.calculate(seqs, 0, seq1.getLength(), result, false); + int width = seq1.getLength(); + ProfilesI result = AAFrequency.calculate(seqs, width, 0, width, + false); // col 0 is 100% C - Profile col = result[0]; + ProfileI col = result.get(0); assertEquals(100f, col.getPercentageIdentity(false)); assertEquals(100f, col.getPercentageIdentity(true)); assertEquals(4, col.getMaxCount()); @@ -53,28 +89,28 @@ public class AAFrequencyTest assertNull(col.getCounts()); // col 1 is 75% A - col = result[1]; + col = result.get(1); assertEquals(75f, col.getPercentageIdentity(false)); assertEquals(100f, col.getPercentageIdentity(true)); assertEquals(3, col.getMaxCount()); assertEquals("A", col.getModalResidue()); // col 2 is 50% G 50% C or 25/25 counting gaps - col = result[2]; + col = result.get(2); assertEquals(25f, col.getPercentageIdentity(false)); assertEquals(50f, col.getPercentageIdentity(true)); assertEquals(1, col.getMaxCount()); assertEquals("CG", col.getModalResidue()); // col 3 is all gaps - col = result[3]; + col = result.get(3); assertEquals(0f, col.getPercentageIdentity(false)); assertEquals(0f, col.getPercentageIdentity(true)); assertEquals(0, col.getMaxCount()); assertEquals("", col.getModalResidue()); // col 4 is 75% T 25% G - col = result[4]; + col = result.get(4); assertEquals(75f, col.getPercentageIdentity(false)); assertEquals(75f, col.getPercentageIdentity(true)); assertEquals(3, col.getMaxCount()); @@ -89,26 +125,27 @@ public class AAFrequencyTest SequenceI seq3 = new Sequence("Seq3", "C--G"); SequenceI seq4 = new Sequence("Seq4", "CA-t"); SequenceI[] seqs = new SequenceI[] { seq1, seq2, seq3, seq4 }; - Profile[] result = new Profile[seq1.getLength()]; + int width = seq1.getLength(); + ProfilesI result = AAFrequency.calculate(seqs, width, 0, width, + true); - AAFrequency.calculate(seqs, 0, seq1.getLength(), result, true); - Profile profile = result[0]; + ProfileI profile = result.get(0); assertEquals(4, profile.getCounts().getCount('C')); assertEquals(4, profile.getHeight()); assertEquals(4, profile.getNonGapped()); - profile = result[1]; + profile = result.get(1); assertEquals(3, profile.getCounts().getCount('A')); assertEquals(4, profile.getHeight()); assertEquals(3, profile.getNonGapped()); - profile = result[2]; + profile = result.get(2); assertEquals(1, profile.getCounts().getCount('C')); assertEquals(1, profile.getCounts().getCount('G')); assertEquals(4, profile.getHeight()); assertEquals(2, profile.getNonGapped()); - profile = result[3]; + profile = result.get(3); assertEquals(3, profile.getCounts().getCount('T')); assertEquals(1, profile.getCounts().getCount('G')); assertEquals(4, profile.getHeight()); @@ -123,15 +160,16 @@ public class AAFrequencyTest SequenceI seq3 = new Sequence("Seq3", "C--G"); SequenceI seq4 = new Sequence("Seq4", "CA-t"); SequenceI[] seqs = new SequenceI[] { seq1, seq2, seq3, seq4 }; - Profile[] result = new Profile[seq1.getLength()]; - // ensure class loaded and initialized - AAFrequency.calculate(seqs, 0, seq1.getLength(), result, true); + // ensure class loaded and initialised + int width = seq1.getLength(); + AAFrequency.calculate(seqs, width, 0, width, true); + int reps = 100000; long start = System.currentTimeMillis(); for (int i = 0; i < reps; i++) { - AAFrequency.calculate(seqs, 0, seq1.getLength(), result, true); + AAFrequency.calculate(seqs, width, 0, width, true); } System.out.println(System.currentTimeMillis() - start); } @@ -153,11 +191,11 @@ public class AAFrequencyTest SequenceI seq3 = new Sequence("Seq3", "C---G"); SequenceI seq4 = new Sequence("Seq4", "CA--t"); SequenceI[] seqs = new SequenceI[] { seq1, seq2, seq3, seq4 }; - Profile[] profiles = new Profile[seq1.getLength()]; - AAFrequency.calculate(seqs, 0, seq1.getLength(), profiles, true); + int width = seq1.getLength(); + ProfilesI profiles = AAFrequency.calculate(seqs, width, 0, width, true); AlignmentAnnotation consensus = new AlignmentAnnotation("Consensus", - "PID", new Annotation[seq1.getLength()]); + "PID", new Annotation[width]); AAFrequency .completeConsensus(consensus, profiles, 0, 5, false, true, 4); @@ -194,11 +232,11 @@ public class AAFrequencyTest SequenceI seq3 = new Sequence("Seq3", "C---G"); SequenceI seq4 = new Sequence("Seq4", "CA--t"); SequenceI[] seqs = new SequenceI[] { seq1, seq2, seq3, seq4 }; - Profile[] profiles = new Profile[seq1.getLength()]; - AAFrequency.calculate(seqs, 0, seq1.getLength(), profiles, true); + int width = seq1.getLength(); + ProfilesI profiles = AAFrequency.calculate(seqs, width, 0, width, true); AlignmentAnnotation consensus = new AlignmentAnnotation("Consensus", - "PID", new Annotation[seq1.getLength()]); + "PID", new Annotation[width]); AAFrequency .completeConsensus(consensus, profiles, 0, 5, true, false, 4); @@ -218,4 +256,272 @@ public class AAFrequencyTest assertEquals("T 75%", ann.description); assertEquals("T", ann.displayCharacter); } + + /** + * Test to include rounding down of a non-zero count to 0% (JAL-3202) + */ + @Test(groups = { "Functional" }) + public void testExtractProfile() + { + /* + * 200 sequences of which 30 gapped (170 ungapped) + * max count 70 for modal residue 'G' + */ + ProfileI profile = new Profile(200, 30, 70, "G"); + ResidueCount counts = new ResidueCount(); + counts.put('G', 70); + counts.put('R', 60); + counts.put('L', 38); + counts.put('H', 2); + profile.setCounts(counts); + + /* + * [0, noOfValues, totalPercent, char1, count1, ...] + * G: 70/170 = 41.2 = 41 + * R: 60/170 = 35.3 = 35 + * L: 38/170 = 22.3 = 22 + * H: 2/170 = 1 + * total (rounded) percentages = 99 + */ + int[] extracted = AAFrequency.extractProfile(profile, true); + int[] expected = new int[] { 0, 4, 99, 'G', 41, 'R', 35, 'L', 22, 'H', + 1 }; + org.testng.Assert.assertEquals(extracted, expected); + + /* + * add some counts of 1; these round down to 0% and should be discarded + */ + counts.put('G', 68); // 68/170 = 40% exactly (percentages now total 98) + counts.put('Q', 1); + counts.put('K', 1); + extracted = AAFrequency.extractProfile(profile, true); + expected = new int[] { 0, 4, 98, 'G', 40, 'R', 35, 'L', 22, 'H', 1 }; + org.testng.Assert.assertEquals(extracted, expected); + + } + + /** + * Tests for the profile calculation where gaps are included i.e. the + * denominator is the total number of sequences in the column + */ + @Test(groups = { "Functional" }) + public void testExtractProfile_countGaps() + { + /* + * 200 sequences of which 30 gapped (170 ungapped) + * max count 70 for modal residue 'G' + */ + ProfileI profile = new Profile(200, 30, 70, "G"); + ResidueCount counts = new ResidueCount(); + counts.put('G', 70); + counts.put('R', 60); + counts.put('L', 38); + counts.put('H', 2); + profile.setCounts(counts); + + /* + * [0, noOfValues, totalPercent, char1, count1, ...] + * G: 70/200 = 35% + * R: 60/200 = 30% + * L: 38/200 = 19% + * H: 2/200 = 1% + * total (rounded) percentages = 85 + */ + int[] extracted = AAFrequency.extractProfile(profile, false); + int[] expected = new int[] { AlignmentAnnotation.SEQUENCE_PROFILE, 4, + 85, 'G', 35, 'R', 30, 'L', 19, 'H', + 1 }; + org.testng.Assert.assertEquals(extracted, expected); + + /* + * add some counts of 1; these round down to 0% and should be discarded + */ + counts.put('G', 68); // 68/200 = 34% + counts.put('Q', 1); + counts.put('K', 1); + extracted = AAFrequency.extractProfile(profile, false); + expected = new int[] { AlignmentAnnotation.SEQUENCE_PROFILE, 4, 84, 'G', + 34, 'R', 30, 'L', 19, 'H', 1 }; + org.testng.Assert.assertEquals(extracted, expected); + + } + + @Test(groups = { "Functional" }) + public void testExtractCdnaProfile() + { + /* + * 200 sequences of which 30 gapped (170 ungapped) + * max count 70 for modal residue 'G' + */ + Hashtable profile = new Hashtable(); + + /* + * cdna profile is {seqCount, ungappedCount, codonCount1, ...codonCount64} + * where 1..64 positions correspond to encoded codons + * see CodingUtils.encodeCodon() + */ + int[] codonCounts = new int[66]; + char[] codon1 = new char[] { 'G', 'C', 'A' }; + char[] codon2 = new char[] { 'c', 'C', 'A' }; + char[] codon3 = new char[] { 't', 'g', 'A' }; + char[] codon4 = new char[] { 'G', 'C', 't' }; + int encoded1 = CodingUtils.encodeCodon(codon1); + int encoded2 = CodingUtils.encodeCodon(codon2); + int encoded3 = CodingUtils.encodeCodon(codon3); + int encoded4 = CodingUtils.encodeCodon(codon4); + codonCounts[2 + encoded1] = 30; + codonCounts[2 + encoded2] = 70; + codonCounts[2 + encoded3] = 9; + codonCounts[2 + encoded4] = 1; + codonCounts[0] = 120; + codonCounts[1] = 110; + profile.put(AAFrequency.PROFILE, codonCounts); + + /* + * [0, noOfValues, totalPercent, char1, count1, ...] + * codon1: 30/110 = 27.2 = 27% + * codon2: 70/110 = 63.6% = 63% + * codon3: 9/110 = 8.1% = 8% + * codon4: 1/110 = 0.9% = 0% should be discarded + * total (rounded) percentages = 98 + */ + int[] extracted = AAFrequency.extractCdnaProfile(profile, true); + int[] expected = new int[] { AlignmentAnnotation.CDNA_PROFILE, 3, 98, + encoded2, 63, encoded1, 27, encoded3, 8 }; + org.testng.Assert.assertEquals(extracted, expected); + } + + @Test(groups = { "Functional" }) + public void testExtractCdnaProfile_countGaps() + { + /* + * 200 sequences of which 30 gapped (170 ungapped) + * max count 70 for modal residue 'G' + */ + Hashtable profile = new Hashtable(); + + /* + * cdna profile is {seqCount, ungappedCount, codonCount1, ...codonCount64} + * where 1..64 positions correspond to encoded codons + * see CodingUtils.encodeCodon() + */ + int[] codonCounts = new int[66]; + char[] codon1 = new char[] { 'G', 'C', 'A' }; + char[] codon2 = new char[] { 'c', 'C', 'A' }; + char[] codon3 = new char[] { 't', 'g', 'A' }; + char[] codon4 = new char[] { 'G', 'C', 't' }; + int encoded1 = CodingUtils.encodeCodon(codon1); + int encoded2 = CodingUtils.encodeCodon(codon2); + int encoded3 = CodingUtils.encodeCodon(codon3); + int encoded4 = CodingUtils.encodeCodon(codon4); + codonCounts[2 + encoded1] = 30; + codonCounts[2 + encoded2] = 70; + codonCounts[2 + encoded3] = 9; + codonCounts[2 + encoded4] = 1; + codonCounts[0] = 120; + codonCounts[1] = 110; + profile.put(AAFrequency.PROFILE, codonCounts); + + /* + * [0, noOfValues, totalPercent, char1, count1, ...] + * codon1: 30/120 = 25% + * codon2: 70/120 = 58.3 = 58% + * codon3: 9/120 = 7.5 = 7% + * codon4: 1/120 = 0.8 = 0% should be discarded + * total (rounded) percentages = 90 + */ + int[] extracted = AAFrequency.extractCdnaProfile(profile, false); + int[] expected = new int[] { AlignmentAnnotation.CDNA_PROFILE, 3, 90, + encoded2, 58, encoded1, 25, encoded3, 7 }; + org.testng.Assert.assertEquals(extracted, expected); + } + + @Test(groups = { "Functional" }) + public void testExtractHMMProfile() + throws MalformedURLException, IOException + { + int[] expected = { 0, 4, 100, 'T', 71, 'C', 12, 'G', 9, 'A', 9 }; + int[] actual = AAFrequency.extractHMMProfile(hmm, 17, false, false); + for (int i = 0; i < actual.length; i++) + { + if (i == 2) + { + assertEquals(actual[i], expected[i]); + } + else + { + assertEquals(actual[i], expected[i]); + } + } + + int[] expected2 = { 0, 4, 100, 'A', 85, 'C', 0, 'G', 0, 'T', 0 }; + int[] actual2 = AAFrequency.extractHMMProfile(hmm, 2, true, false); + for (int i = 0; i < actual2.length; i++) + { + if (i == 2) + { + assertEquals(actual[i], expected[i]); + } + else + { + assertEquals(actual[i], expected[i]); + } + } + + assertNull(AAFrequency.extractHMMProfile(null, 98978867, true, false)); + } + + @Test(groups = { "Functional" }) + public void testGetAnalogueCount() + { + /* + * 'T' in column 0 has emission probability 0.7859, scales to 7859 + */ + int count = AAFrequency.getAnalogueCount(hmm, 0, 'T', false, false); + assertEquals(7859, count); + + /* + * same with 'use info height': value is multiplied by log ratio + * log(value / background) / log(2) = log(0.7859/0.25)/0.693 + * = log(3.1)/0.693 = 1.145/0.693 = 1.66 + * so value becomes 1.2987 and scales up to 12987 + */ + count = AAFrequency.getAnalogueCount(hmm, 0, 'T', false, true); + assertEquals(12987, count); + + /* + * 'G' in column 20 has emission probability 0.75457, scales to 7546 + */ + count = AAFrequency.getAnalogueCount(hmm, 20, 'G', false, false); + assertEquals(7546, count); + + /* + * 'G' in column 1077 has emission probability 0.0533, here + * ignored (set to 0) since below background of 0.25 + */ + count = AAFrequency.getAnalogueCount(hmm, 1077, 'G', true, false); + assertEquals(0, count); + } + + @Test(groups = { "Functional" }) + public void testCompleteInformation() + { + ProfileI prof1 = new Profile(1, 0, 100, "A"); + ProfileI prof2 = new Profile(1, 0, 100, "-"); + + ProfilesI profs = new Profiles(new ProfileI[] { prof1, prof2 }); + Annotation ann1 = new Annotation(6.5f); + Annotation ann2 = new Annotation(0f); + Annotation[] annots = new Annotation[] { ann1, ann2 }; + SequenceI seq = new Sequence("", "AA", 0, 0); + seq.setHMM(hmm); + AlignmentAnnotation annot = new AlignmentAnnotation("", "", annots); + annot.setSequenceRef(seq); + AAFrequency.completeInformation(annot, profs, 0, 1); + float ic = annot.annotations[0].value; + assertEquals(0.91532f, ic, 0.0001f); + ic = annot.annotations[1].value; + assertEquals(0f, ic, 0.0001f); + int i = 0; + } }