Examples with EMDataSet org.baderlab.csplugins.enrichmentmap.model.EMDataSet used on opensource projects

Example 21 with EMDataSet

use of org.baderlab.csplugins.enrichmentmap.model.EMDataSet in project EnrichmentMapApp by BaderLab.

the class HeatMapRanksTest method testLeadingEdge.

@Test
public void testLeadingEdge(EnrichmentMapManager emManager) throws Exception {
    final String geneSetName = "ENVELOPE%GO%GO:0031975";
    final int leadingEdgeSize = 170;
    // Sanity test
    EnrichmentMap map = emManager.getAllEnrichmentMaps().values().iterator().next();
    EMDataSet dataset = map.getDataSet(LegacySupport.DATASET1);
    GeneSet gs = dataset.getGeneSetsOfInterest().getGeneSets().get(geneSetName);
    assertNotNull(gs);
    // Run the ranking
    RankingOption rankingOption = new GSEALeadingEdgeRankingOption(dataset, geneSetName, Ranking.GSEARanking);
    Map<Integer, RankValue> ranks = rankingOption.computeRanking(gs.getGenes()).get().get();
    assertEquals(454, ranks.size());
    // Convert to useful collections
    Map<RankValue, Integer> rankToGeneId = HashBiMap.create(ranks).inverse();
    List<RankValue> sortedRanks = ranks.values().stream().sorted().collect(Collectors.toList());
    // Test leading edge
    for (int i = 0; i < sortedRanks.size(); i++) {
        RankValue v = sortedRanks.get(i);
        assertTrue(v.isSignificant() == i < leadingEdgeSize);
    }
    // Test genes are the same
    List<Integer> expectedGeneOrder = getGeneOrderFromFile(map, PATH + "gene_order_leading_edge.txt");
    List<Integer> actualGeneOrder = sortedRanks.stream().map(rankToGeneId::get).collect(Collectors.toList());
    assertEquals(expectedGeneOrder, actualGeneOrder);
}

Example 22 with EMDataSet

use of org.baderlab.csplugins.enrichmentmap.model.EMDataSet in project EnrichmentMapApp by BaderLab.

the class FileReaderTest method testExpression1ReaderRnk.

@Test
public void testExpression1ReaderRnk(Provider<EnrichmentMapParameters> empFactory) throws Exception {
    //load the test expression file
    String testDataFileName = "src/test/resources/org/baderlab/csplugins/enrichmentmap/ExpressionTestFile.rnk";
    //create a new instance of the parameters
    EnrichmentMapParameters params = empFactory.get();
    //set expression file name 
    params.getFiles().get(LegacySupport.DATASET1).setExpressionFileName(testDataFileName);
    //Create a new Enrichment map
    EnrichmentMap map = new EnrichmentMap(params.getCreationParameters(), serviceRegistrar);
    //get the default dataset
    Method method = EnrichmentMapParameters.stringToMethod(params.getMethod());
    DataSetFiles files = params.getFiles().get(LegacySupport.DATASET1);
    EMDataSet dataset = map.createDataSet(LegacySupport.DATASET1, method, files);
    //make sure that the genes are empty
    assertEquals(0, map.getNumberOfGenes());
    //add the gene to the master list of genes
    map.addGene("GLS");
    map.addGene("PSMA1");
    map.addGene("ZP1");
    map.addGene("ZYX");
    //make sure all four genes have been associated
    assertEquals(4, map.getNumberOfGenes());
    //load expression file
    ExpressionFileReaderTask task = new ExpressionFileReaderTask(dataset);
    task.run(taskMonitor);
    //There was one more gene in the expression file that wasn't in the set of genes
    //make sure it was was added
    assertEquals(4, map.getNumberOfGenes());
    assertEquals(4, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getNumGenes());
    assertEquals(3, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getNumConditions());
    assertEquals(0.47536945, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getMinExpression(), 0.0);
    assertEquals(0.5418719, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getMaxExpression(), 0.0);
}

Example 23 with EMDataSet

use of org.baderlab.csplugins.enrichmentmap.model.EMDataSet in project EnrichmentMapApp by BaderLab.

the class FileReaderTest method testExpression1ReaderEDBRnk.

@Test
public void testExpression1ReaderEDBRnk(Provider<EnrichmentMapParameters> empFactory) throws Exception {
    //load the test expression file
    String testDataFileName = "src/test/resources/org/baderlab/csplugins/enrichmentmap/ExpressionTestFile_edbrnk.rnk";
    //create a new instance of the parameters
    EnrichmentMapParameters params = empFactory.get();
    //set gmt file name 
    params.getFiles().get(LegacySupport.DATASET1).setExpressionFileName(testDataFileName);
    //Create a new Enrichment map
    EnrichmentMap map = new EnrichmentMap(params.getCreationParameters(), serviceRegistrar);
    //get the default dataset
    Method method = EnrichmentMapParameters.stringToMethod(params.getMethod());
    DataSetFiles files = params.getFiles().get(LegacySupport.DATASET1);
    EMDataSet dataset = map.createDataSet(LegacySupport.DATASET1, method, files);
    //make sure that the genes are empty
    assertEquals(0, map.getNumberOfGenes());
    //add the gene to the master list of genes
    map.addGene("GLS");
    map.addGene("PSMA1");
    map.addGene("ZP1");
    map.addGene("ZYX");
    //make sure all four genes have been associated
    assertEquals(4, map.getNumberOfGenes());
    //load expression file
    ExpressionFileReaderTask task = new ExpressionFileReaderTask(dataset);
    task.run(taskMonitor);
    //There was one more gene in the expression file that wasn't in the set of genes
    //make sure it was was added
    assertEquals(4, map.getNumberOfGenes());
    assertEquals(4, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getNumGenes());
    assertEquals(3, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getNumConditions());
    assertEquals(0.47536945, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getMinExpression(), 0.0);
    assertEquals(0.5418719, map.getDataSet(LegacySupport.DATASET1).getExpressionSets().getMaxExpression(), 0.0);
}

Example 24 with EMDataSet

use of org.baderlab.csplugins.enrichmentmap.model.EMDataSet in project EnrichmentMapApp by BaderLab.

the class FileReaderTest method testGSEAEnrichmentsReader.

//test GSEA enrichment results reader
@Test
public void testGSEAEnrichmentsReader(Provider<EnrichmentMapParameters> empFactory) throws Exception {
    //load the test enrichment files - GSEA creates two enrichment results files.
    String testDataFileName = "src/test/resources/org/baderlab/csplugins/enrichmentmap/GSEA_enrichments1.xls";
    String testDataFileName2 = "src/test/resources/org/baderlab/csplugins/enrichmentmap/GSEA_enrichments2.xls";
    //create a new instance of the parameters
    EnrichmentMapParameters params = empFactory.get();
    //set enrichment file name 
    params.getFiles().get(LegacySupport.DATASET1).setEnrichmentFileName1(testDataFileName);
    params.getFiles().get(LegacySupport.DATASET1).setEnrichmentFileName2(testDataFileName2);
    //Create a new Enrichment map
    EnrichmentMap map = new EnrichmentMap(params.getCreationParameters(), serviceRegistrar);
    //get the default dataset
    Method method = EnrichmentMapParameters.stringToMethod(params.getMethod());
    DataSetFiles files = params.getFiles().get(LegacySupport.DATASET1);
    EMDataSet dataset = map.createDataSet(LegacySupport.DATASET1, method, files);
    ParseGSEAEnrichmentResults task = new ParseGSEAEnrichmentResults(dataset);
    task.run(taskMonitor);
    //Get the enrichment
    Map<String, EnrichmentResult> enrichments = map.getDataSet(LegacySupport.DATASET1).getEnrichments().getEnrichments();
    assertEquals(40, enrichments.size());
    //Check the contents of some of the genesets
    // example from file 1 (ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612)
    //check p-values
    assertEquals(0.0, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getPvalue(), 0.0);
    //check fdr value
    assertEquals(0.086938426, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getFdrqvalue(), 0.0);
    //check ES value
    assertEquals(0.6854155, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getES(), 0.0);
    //check NES
    assertEquals(2.1194055, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getNES(), 0.0);
    //check ranks at max
    assertEquals(836, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getRankAtMax());
    //check size
    assertEquals(27, ((GSEAResult) enrichments.get("ANTIGEN PROCESSING AND PRESENTATION%KEGG%HSA04612")).getGsSize());
    // example from file 2 (EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143)
    //check p-values
    assertEquals(0.040152963, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getPvalue(), 0.0);
    //check fdr value
    assertEquals(1.0, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getFdrqvalue(), 0.0);
    //check ES value
    assertEquals(-0.49066687, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getES(), 0.0);
    //check NES
    assertEquals(-1.477554, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getNES(), 0.0);
    //check ranks at max
    assertEquals(1597, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getRankAtMax());
    //check size
    assertEquals(17, ((GSEAResult) enrichments.get("EMBRYONIC HEART TUBE MORPHOGENESIS%GO%GO:0003143")).getGsSize());
}

Example 25 with EMDataSet

use of org.baderlab.csplugins.enrichmentmap.model.EMDataSet in project EnrichmentMapApp by BaderLab.

the class LegacySessionLoadTest method test_1_LoadedLegacyData.

@Test
@SessionFile("em_session_2.2.cys")
public void test_1_LoadedLegacyData() throws Exception {
    EnrichmentMap map = getEnrichmentMap();
    assertEquals("EM1_Enrichment Map", map.getName());
    CyNetwork network = networkManager.getNetwork(map.getNetworkID());
    assertNotNull(network);
    assertEquals(1, map.getDataSetCount());
    assertEquals(14067, map.getNumberOfGenes());
    assertEquals(14067, map.getAllGenes().size());
    // Number of edges: 3339 - that's how many geneset similarity objects there should be!!!
    CyTable edgeTable = network.getDefaultEdgeTable();
    assertEquals(3339, edgeTable.getRowCount());
    EMCreationParameters params = map.getParams();
    String prefix = params.getAttributePrefix();
    assertEquals("EM1_", prefix);
    assertEquals(0.5, params.getCombinedConstant(), 0.0);
    assertFalse(params.isEMgmt());
    assertEquals("Geneset_Overlap", params.getEnrichmentEdgeType());
    assertTrue(params.isFDR());
    assertEquals(GreatFilter.HYPER, params.getGreatFilter());
    assertEquals(0.005, params.getPvalue(), 0.0);
    assertEquals(1.0, params.getPvalueMin(), 0.0);
    assertEquals(0.1, params.getQvalue(), 0.0);
    assertEquals(1.0, params.getQvalueMin(), 0.0);
    assertEquals(0.5, params.getSimilarityCutoff(), 0.0);
    assertEquals(SimilarityMetric.OVERLAP, params.getSimilarityMetric());
    //		assertFalse(params.isDistinctExpressionSets());
    String geneset1 = "RESOLUTION OF SISTER CHROMATID COHESION%REACTOME%REACT_150425.2";
    String geneset2 = "CHROMOSOME, CENTROMERIC REGION%GO%GO:0000775";
    Collection<CyRow> rows = edgeTable.getMatchingRows(CyNetwork.NAME, geneset1 + " (Geneset_Overlap) " + geneset2);
    assertEquals(1, rows.size());
    CyRow row = rows.iterator().next();
    assertEquals("Geneset_Overlap", row.get(CyEdge.INTERACTION, String.class));
    assertEquals(0.6097560975609756, EMStyleBuilder.Columns.EDGE_SIMILARITY_COEFF.get(row, prefix), 0.0);
    EMDataSet dataset = map.getDataSet("Dataset 1");
    assertNotNull(dataset);
    assertSame(map, dataset.getMap());
    assertEquals(Method.GSEA, dataset.getMethod());
    assertEquals(12653, dataset.getDataSetGenes().size());
    assertEquals(389, dataset.getGeneSetsOfInterest().getGeneSets().size());
    //		assertEquals(17259, dataset.getSetofgenesets().getGenesets().size()); // MKTODO why? what is this used for
    assertEndsWith(dataset.getSetOfGeneSets().getFilename(), "Human_GO_AllPathways_no_GO_iea_April_15_2013_symbol.gmt");
    for (long suid : dataset.getNodeSuids()) {
        assertNotNull(network.getNode(suid));
    }
    GeneSet geneset = dataset.getGeneSetsOfInterest().getGeneSets().get("NCRNA PROCESSING%GO%GO:0034470");
    assertEquals(88, geneset.getGenes().size());
    assertEquals("NCRNA PROCESSING%GO%GO:0034470", geneset.getName());
    assertEquals("ncRNA processing", geneset.getDescription());
    assertEquals(Optional.of("GO"), geneset.getSource());
    SetOfEnrichmentResults enrichments = dataset.getEnrichments();
    assertEquals(4756, enrichments.getEnrichments().size());
    assertEndsWith(enrichments.getFilename1(), "gsea_report_for_ES12_1473194913081.xls");
    assertEndsWith(enrichments.getFilename2(), "gsea_report_for_NT12_1473194913081.xls");
    assertEquals("ES12", enrichments.getPhenotype1());
    assertEquals("NT12", enrichments.getPhenotype2());
    EnrichmentResult result = enrichments.getEnrichments().get("RIBONUCLEOSIDE TRIPHOSPHATE BIOSYNTHETIC PROCESS%GO%GO:0009201");
    assertTrue(result instanceof GSEAResult);
    GSEAResult gseaResult = (GSEAResult) result;
    assertEquals("RIBONUCLEOSIDE TRIPHOSPHATE BIOSYNTHETIC PROCESS%GO%GO:0009201", gseaResult.getName());
    assertEquals(0.42844063, gseaResult.getES(), 0.0);
    assertEquals(0.45225498, gseaResult.getFdrqvalue(), 0.0);
    assertEquals(1.0, gseaResult.getFwerqvalue(), 0.0);
    assertEquals(23, gseaResult.getGsSize());
    assertEquals(1.1938541, gseaResult.getNES(), 0.0);
    assertEquals(0.2457786, gseaResult.getPvalue(), 0.0);
    assertEquals(4689, gseaResult.getRankAtMax());
    assertEquals(Optional.of("GO"), gseaResult.getSource());
    GeneExpressionMatrix expressions = dataset.getExpressionSets();
    assertEquals(20326, expressions.getExpressionUniverse());
    assertEquals(3.686190609, expressions.getClosesttoZero(), 0.0);
    //		assertEndsWith(expressions.getFilename(), "MCF7_ExprMx_v2_names.gct");
    assertEquals(15380.42388, expressions.getMaxExpression(), 0.0);
    assertEquals(3.686190609, expressions.getMinExpression(), 0.0);
    assertEquals(20, expressions.getNumConditions());
    assertEquals(12653, expressions.getExpressionMatrix().size());
    assertEquals(12653, expressions.getExpressionMatrix_rowNormalized().size());
    GeneExpression expression = expressions.getExpressionMatrix().get(0);
    assertEquals("MOCOS", expression.getName());
    assertEquals("MOCOS (molybdenum cofactor sulfurase)", expression.getDescription());
    assertEquals(18, expression.getExpression().length);
    Ranking ranking = expressions.getRanks().get("GSEARanking");
    assertEquals(12653, ranking.getAllRanks().size());
    assertEquals(12653, ranking.getRanking().size());
    Rank rank = ranking.getRanking().get(0);
    assertEquals("MOCOS", rank.getName());
    assertEquals(1238, rank.getRank().intValue());
    assertEquals(0.54488367, rank.getScore(), 0.0);
    DataSetFiles files = dataset.getDataSetFiles();
    assertEndsWith(files.getClassFile(), "ES_NT.cls");
    assertEndsWith(files.getEnrichmentFileName1(), "gsea_report_for_ES12_1473194913081.xls");
    assertEndsWith(files.getEnrichmentFileName2(), "gsea_report_for_NT12_1473194913081.xls");
    //		assertEndsWith(files.getExpressionFileName(), "MCF7_ExprMx_v2_names.gct");
    assertEndsWith(files.getGMTFileName(), "Human_GO_AllPathways_no_GO_iea_April_15_2013_symbol.gmt");
    assertEndsWith(files.getGseaHtmlReportFile(), "estrogen_treatment_12hr_gsea_enrichment_results.Gsea.1473194913081/index.html");
    assertEndsWith(files.getRankedFile(), "ranked_gene_list_ES12_versus_NT12_1473194913081.xls");
    assertEquals("ES12", files.getPhenotype1());
    assertEquals("NT12", files.getPhenotype2());
}