From 06827cdbf16226b2a625f6a94bde942b0e74b770 Mon Sep 17 00:00:00 2001 From: gmungoc Date: Wed, 27 Feb 2019 12:31:04 +0000 Subject: [PATCH] JAL-3202 discard zero percentages in extracted profile --- src/jalview/analysis/AAFrequency.java | 54 ++++++-- src/jalview/renderer/AnnotationRenderer.java | 14 +- test/jalview/analysis/AAFrequencyTest.java | 183 ++++++++++++++++++++++++++ 3 files changed, 235 insertions(+), 16 deletions(-) diff --git a/src/jalview/analysis/AAFrequency.java b/src/jalview/analysis/AAFrequency.java index e4f2dfa..a1b0325 100755 --- a/src/jalview/analysis/AAFrequency.java +++ b/src/jalview/analysis/AAFrequency.java @@ -398,7 +398,7 @@ public class AAFrequency * contains * * 
-   *    [profileType, numberOfValues, nonGapCount, charValue1, percentage1, charValue2, percentage2, ...]
+   *    [profileType, numberOfValues, totalPercent, charValue1, percentage1, charValue2, percentage2, ...]
    * in descending order of percentage value
    *
* @@ -411,7 +411,6 @@ public class AAFrequency */ public static int[] extractProfile(ProfileI profile, boolean ignoreGaps) { - int[] rtnval = new int[64]; ResidueCount counts = profile.getCounts(); if (counts == null) { @@ -422,7 +421,6 @@ public class AAFrequency char[] symbols = symbolCounts.symbols; int[] values = symbolCounts.values; QuickSort.sort(values, symbols); - int nextArrayPos = 2; int totalPercentage = 0; final int divisor = ignoreGaps ? profile.getNonGapped() : profile.getHeight(); @@ -430,21 +428,44 @@ public class AAFrequency /* * traverse the arrays in reverse order (highest counts first) */ + int[] result = new int[3 + 2 * symbols.length]; + int nextArrayPos = 3; + int nonZeroCount = 0; + for (int i = symbols.length - 1; i >= 0; i--) { int theChar = symbols[i]; int charCount = values[i]; - - rtnval[nextArrayPos++] = theChar; final int percentage = (charCount * 100) / divisor; - rtnval[nextArrayPos++] = percentage; + if (percentage == 0) + { + /* + * this count (and any remaining) round down to 0% - discard + */ + break; + } + nonZeroCount++; + result[nextArrayPos++] = theChar; + result[nextArrayPos++] = percentage; totalPercentage += percentage; } - rtnval[0] = symbols.length; - rtnval[1] = totalPercentage; - int[] result = new int[rtnval.length + 1]; + + /* + * truncate array if any zero values were discarded + */ + if (nonZeroCount < symbols.length) + { + int[] tmp = new int[3 + 2 * nonZeroCount]; + System.arraycopy(result, 0, tmp, 0, tmp.length); + result = tmp; + } + + /* + * fill in 'header' values + */ result[0] = AlignmentAnnotation.SEQUENCE_PROFILE; - System.arraycopy(rtnval, 0, result, 1, rtnval.length); + result[1] = nonZeroCount; + result[2] = totalPercentage; return result; } @@ -454,7 +475,7 @@ public class AAFrequency * contains * * 
-   *    [profileType, numberOfValues, totalCount, charValue1, percentage1, charValue2, percentage2, ...]
+   *    [profileType, numberOfValues, totalPercentage, charValue1, percentage1, charValue2, percentage2, ...]
    * in descending order of percentage value, where the character values encode codon triplets
    *
* @@ -492,9 +513,16 @@ public class AAFrequency { break; // nothing else of interest here } + final int percentage = codonCount * 100 / divisor; + if (percentage == 0) + { + /* + * this (and any remaining) values rounded down to 0 - discard + */ + break; + } distinctValuesCount++; result[j++] = codons[i]; - final int percentage = codonCount * 100 / divisor; result[j++] = percentage; totalPercentage += percentage; } @@ -531,7 +559,7 @@ public class AAFrequency for (int col = 0; col < cols; col++) { // todo would prefer a Java bean for consensus data - Hashtable columnHash = new Hashtable(); + Hashtable columnHash = new Hashtable<>(); // #seqs, #ungapped seqs, counts indexed by (codon encoded + 1) int[] codonCounts = new int[66]; codonCounts[0] = alignment.getSequences().size(); diff --git a/src/jalview/renderer/AnnotationRenderer.java b/src/jalview/renderer/AnnotationRenderer.java index adca17e..c849e8c 100644 --- a/src/jalview/renderer/AnnotationRenderer.java +++ b/src/jalview/renderer/AnnotationRenderer.java @@ -471,8 +471,10 @@ public class AnnotationRenderer .getAlignmentStrucConsensusAnnotation(); final AlignmentAnnotation complementConsensusAnnot = av .getComplementConsensusAnnotation(); - boolean renderHistogram = true, renderProfile = true, - normaliseProfile = false, isRNA = rna; + boolean renderHistogram = true; + boolean renderProfile = false; + boolean normaliseProfile = false; + boolean isRNA = rna; BitSet graphGroupDrawn = new BitSet(); int charOffset = 0; // offset for a label @@ -1426,7 +1428,13 @@ public class AnnotationRenderer ht += scl; // next profl[] position is profile % for the character(s) - scl = htn * scale * profl[c++]; + int percent = profl[c++]; + if (percent == 0) + { + // failsafe in case a count rounds down to 0% + continue; + } + scl = htn * scale * percent; lm = ofont.getLineMetrics(dc, 0, 1, g.getFontMetrics().getFontRenderContext()); Font font = ofont.deriveFont(AffineTransform diff --git a/test/jalview/analysis/AAFrequencyTest.java b/test/jalview/analysis/AAFrequencyTest.java index 75fb39e..93c95ce 100644 --- a/test/jalview/analysis/AAFrequencyTest.java +++ b/test/jalview/analysis/AAFrequencyTest.java @@ -25,12 +25,16 @@ import static org.testng.AssertJUnit.assertNull; import jalview.datamodel.AlignmentAnnotation; import jalview.datamodel.Annotation; +import jalview.datamodel.Profile; import jalview.datamodel.ProfileI; import jalview.datamodel.ProfilesI; +import jalview.datamodel.ResidueCount; import jalview.datamodel.Sequence; import jalview.datamodel.SequenceI; import jalview.gui.JvOptionPane; +import java.util.Hashtable; + import org.testng.annotations.BeforeClass; import org.testng.annotations.Test; @@ -232,4 +236,183 @@ 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); + } } -- 1.7.10.2