|
| 1 | +package org.biojava.nbio.genome; |
| 2 | + |
| 3 | +import com.google.common.collect.Lists; |
| 4 | +import com.google.common.collect.Range; |
| 5 | +import junit.framework.TestCase; |
| 6 | +import org.biojava.nbio.genome.parsers.genename.GeneChromosomePosition; |
| 7 | +import org.biojava.nbio.genome.parsers.genename.GeneChromosomePositionParser; |
| 8 | +import org.biojava.nbio.genome.util.ChromosomeMappingTools; |
| 9 | +import org.junit.Test; |
| 10 | + |
| 11 | +import java.io.InputStream; |
| 12 | +import java.net.URL; |
| 13 | +import java.util.List; |
| 14 | +import java.util.zip.GZIPInputStream; |
| 15 | + |
| 16 | +/** |
| 17 | + * Created by andreas on 7/19/16. |
| 18 | + */ |
| 19 | +public class TestGenomeMapping extends TestCase{ |
| 20 | + |
| 21 | + private static final String geneChromosomeFile = "http://cdn.rcsb.org/gene/hg38/geneChromosome38.tsf.gz"; |
| 22 | + |
| 23 | + private List<GeneChromosomePosition> gcps = null; |
| 24 | + |
| 25 | + @Override |
| 26 | + protected void setUp() throws Exception { |
| 27 | + super.setUp(); |
| 28 | + InputStream input = new GZIPInputStream(new URL(geneChromosomeFile).openStream()); |
| 29 | + gcps = GeneChromosomePositionParser.getChromosomeMappings(input); |
| 30 | + |
| 31 | + |
| 32 | + } |
| 33 | + |
| 34 | + |
| 35 | + @Test |
| 36 | + public void testAK1() { |
| 37 | + String geneName = "AK1"; |
| 38 | + |
| 39 | + assertNotNull(gcps); |
| 40 | + assertTrue("Problems with downloading refFlat file from UCSC browser ", gcps.size() > 100); |
| 41 | + |
| 42 | + int uniProtLength = 194; |
| 43 | + |
| 44 | + try { |
| 45 | + |
| 46 | + for (GeneChromosomePosition pos : gcps) { |
| 47 | + |
| 48 | + //System.out.println(pos.getGeneName()); |
| 49 | + if (!pos.getGeneName().equals(geneName)) |
| 50 | + continue; |
| 51 | + |
| 52 | + /// there are three alternative transcripts for AK1. |
| 53 | + // we are just testing one here: |
| 54 | + |
| 55 | + if ( ! pos.getGenebankId().equals("NM_000476")) |
| 56 | + continue; |
| 57 | + |
| 58 | + assertTrue(pos.getGeneName().equals(geneName)); |
| 59 | + assertTrue(pos.getOrientation().equals('-')); |
| 60 | + assertTrue(pos.getChromosome().equals("chr9")); |
| 61 | + |
| 62 | + List<Range<Integer>> cdsranges = ChromosomeMappingTools.getCDSExonRanges(pos); |
| 63 | + |
| 64 | + validateExon(0,0,7, cdsranges ); |
| 65 | + validateExon(1,7,43, cdsranges ); |
| 66 | + validateExon(2,43,207, cdsranges ); |
| 67 | + validateExon(3,207,324, cdsranges ); |
| 68 | + validateExon(4,324,516, cdsranges ); |
| 69 | + validateExon(5,516,585, cdsranges ); |
| 70 | + |
| 71 | + |
| 72 | + int cdslength = ChromosomeMappingTools.getCDSLength(pos); |
| 73 | + |
| 74 | + assertTrue("CDS length should be 582, but is " + cdslength, cdslength == (uniProtLength *3)); |
| 75 | + |
| 76 | + List<Range<Integer>> chromranges = ChromosomeMappingTools.getChromosomalRangesForCDS(pos); |
| 77 | + |
| 78 | + // we are reverse strand. reverse the order |
| 79 | + chromranges = Lists.reverse(chromranges); |
| 80 | + |
| 81 | + assertTrue(chromranges.size() == 6); |
| 82 | + |
| 83 | + // compare with https://www.ncbi.nlm.nih.gov/CCDS/CcdsBrowse.cgi?REQUEST=CCDS&DATA=CCDS6881 |
| 84 | + validateExon(0,127868008,127868076, chromranges ); |
| 85 | + validateExon(1,127868320,127868512, chromranges ); |
| 86 | + validateExon(2,127871822,127871939, chromranges ); |
| 87 | + validateExon(3,127872689,127872853, chromranges ); |
| 88 | + validateExon(4,127873025,127873061, chromranges ); |
| 89 | + validateExon(5,127874610,127874617, chromranges ); |
| 90 | + |
| 91 | + } |
| 92 | + } catch (Exception e) { |
| 93 | + fail(e.getMessage()); |
| 94 | + } |
| 95 | + } |
| 96 | + |
| 97 | + @Test |
| 98 | + public void testHBA(){ |
| 99 | + |
| 100 | + String geneName = "HBA1"; |
| 101 | + assertNotNull(gcps); |
| 102 | + |
| 103 | + assertTrue("Problems with downloading refFlat file from UCSC browser ", gcps.size() > 100); |
| 104 | + |
| 105 | + try { |
| 106 | + |
| 107 | + for ( GeneChromosomePosition pos : gcps){ |
| 108 | + |
| 109 | + //System.out.println(pos.getGeneName()); |
| 110 | + if ( ! pos.getGeneName().equals(geneName)) |
| 111 | + continue; |
| 112 | + |
| 113 | + assertTrue(pos.getGeneName().equals("HBA1")); |
| 114 | + assertTrue(pos.getGenebankId().equals("NM_000558")); |
| 115 | + assertTrue(pos.getChromosome().equals("chr16")); |
| 116 | + assertTrue(pos.getTranscriptionStart().equals(176650)); |
| 117 | + assertTrue(pos.getTranscriptionEnd().equals(177522)); |
| 118 | + assertTrue(pos.getOrientation().equals('+')); |
| 119 | + |
| 120 | + List<Range<Integer>> cdsranges = ChromosomeMappingTools.getCDSExonRanges(pos); |
| 121 | + |
| 122 | + assertTrue(cdsranges.size() == 3); |
| 123 | + |
| 124 | + validateExon(0,0,95,cdsranges); |
| 125 | + validateExon(1,95,300,cdsranges); |
| 126 | + validateExon(2,300,429,cdsranges); |
| 127 | + |
| 128 | + |
| 129 | + List<Range<Integer>> chromranges = ChromosomeMappingTools.getChromosomalRangesForCDS(pos); |
| 130 | + |
| 131 | + validateExon(0,176716,176811, chromranges ); |
| 132 | + validateExon(1,176928,177133, chromranges ); |
| 133 | + validateExon(2,177282,177411, chromranges ); |
| 134 | + |
| 135 | + |
| 136 | + } |
| 137 | + } catch (Exception e){ |
| 138 | + fail(e.getMessage()); |
| 139 | + } |
| 140 | + |
| 141 | + |
| 142 | + } |
| 143 | + |
| 144 | + private void validateExon(int exonNr, int start, int stop, List<Range<Integer>> cdsranges) { |
| 145 | + |
| 146 | + Range exon = cdsranges.get(exonNr); |
| 147 | + assertTrue("Exon " + exonNr + " boundary "+ exon.lowerEndpoint() + " does not match " +start , exon.lowerEndpoint().equals(start)); |
| 148 | + assertTrue("Exon " + exonNr + " boundary " + exon.upperEndpoint() + " does not match " + stop, exon.upperEndpoint().equals(stop)); |
| 149 | + |
| 150 | + } |
| 151 | +} |
0 commit comments