Set description properly for DAVID results. Fixes #380

Author: mikekucera

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/model/EMCreationParameters.java

@@ -37,6 +37,7 @@ public static enum EdgeStrategy {
 
 	private GreatFilter greatFilter = GreatFilter.HYPER;
 	private boolean fdr = false;
+	private boolean david = false;
 	private boolean emgmt = false;
 	private double qvalueMin = 1.0;
 	private double pvalueMin = 1.0;
@@ -171,6 +172,14 @@ public void setFDR(boolean fdr) {
 		this.fdr = fdr;
 	}
 
+	public void setDavid(boolean david) {
+		this.david = david;
+	}
+	
+	public boolean isDavid() {
+		return david;
+	}
+	
 	public boolean isEMgmt() {
 		return emgmt;
 	}
@@ -269,7 +278,7 @@ public String toString() {
 				+ ", qValueColumnNames=" + qValueColumnNames + ", similarityCutoffColumnNames="
 				+ similarityCutoffColumnNames + "]";
 	}
-	
-	
+
+
 
 }

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/model/GeneSet.java

@@ -59,6 +59,7 @@ public class GeneSet {
 	private final Optional<String> simpleName;
 	private final Optional<String> source;
 	private final Optional<String> datasourceId;
+	private final Optional<String> davidCategory;
 
 
 	public GeneSet(String name, String description, Set<Integer> genes) {
@@ -67,6 +68,7 @@ public GeneSet(String name, String description, Set<Integer> genes) {
 		this.genes = ImmutableSet.copyOf(genes);
 		this.simpleName = Optional.empty();
 		this.datasourceId = Optional.empty();
+		this.davidCategory = Optional.empty();
 
 		// Baderlab geneset names
 		String[] name_tokens = name.split("%");
@@ -76,20 +78,33 @@ public GeneSet(String name, String description, Set<Integer> genes) {
 			this.source = Optional.empty();
 	}
 
-	
-	GeneSet intersectionWith(Set<Integer> expressionGenes) {
-		Set<Integer> intersection = new HashSet<>(genes);
-		intersection.retainAll(expressionGenes);
-		return new GeneSet(name, description, intersection, simpleName.orElse(null), source.orElse(null), datasourceId.orElse(null));
-	}
-	
-	public GeneSet(String name, String description, Set<Integer> genes, String simpleName, String datasource, String datasourceId) {
+	private GeneSet(String name, String description, Set<Integer> genes, 
+			String simpleName, String datasource, String datasourceId, String davidCategory
+	) {
 		this.name = name;
 		this.description = description;
 		this.genes = ImmutableSet.copyOf(genes);
 		this.simpleName = Optional.ofNullable(simpleName);
 		this.source = Optional.ofNullable(datasource);
 		this.datasourceId = Optional.ofNullable(datasourceId);
+		this.davidCategory = Optional.ofNullable(davidCategory);
+	}
+	
+	
+	public static GeneSet createBaderLab(String name, String description, Set<Integer> genes, String simpleName, String datasource, String datasourceId) {
+		return new GeneSet(name, description, genes, simpleName, datasource, datasourceId, null);
+	}
+	
+	public static GeneSet createDavid(String name, String description, Set<Integer> genes, String davidCategory) {
+		return new GeneSet(name, description, genes, null, null, null, davidCategory);
+	}
+	
+	
+	protected GeneSet intersectionWith(Set<Integer> expressionGenes) {
+		Set<Integer> intersection = new HashSet<>(genes);
+		intersection.retainAll(expressionGenes);
+		return new GeneSet(name, description, intersection, 
+				simpleName.orElse(null), source.orElse(null), datasourceId.orElse(null), davidCategory.orElse(null));
 	}
 
 
@@ -117,6 +132,10 @@ public Optional<String> getSimpleName() {
 		return simpleName;
 	}
 
+	public Optional<String> getDavidCategory() {
+		return davidCategory;
+	}
+	
 	public String getLabel() {
 		if(simpleName.isPresent())
 			return simpleName.get();

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/parsers/GMTFileReaderTask.java

@@ -191,7 +191,7 @@ private GeneSet readBaderlabGeneSet(EnrichmentMap map, String line) {
 					builder.add(hash);
 			}
 
-			return new GeneSet(name, description, builder.build(), simpleName, datasource, id);
+			return GeneSet.createBaderLab(name, description, builder.build(), simpleName, datasource, id);
 		}
 		return null;
 	}

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/parsers/ParseDavidEnrichmentResults.java

@@ -76,13 +76,13 @@ public void run(TaskMonitor taskMonitor) throws IOException {
 			double FDRqvalue = 1.0;
 			GenericResult result;
 			int gs_size = 0;
-			double NES = 1.0;
 
 			//The second column of the file is the name of the geneset
-			final String name = tokens[1].toUpperCase().trim();
+			String description = tokens[1].trim();
+			String name = tokens[1].toUpperCase().trim();
 
-			//the first column of the file is the description
-			final String description = tokens[0].toUpperCase();
+			//the first column of the file is the category
+			final String category = tokens[0].toUpperCase();
 
 			//when there are two different species it is possible that the gene set could
 			//already exist in the set of genesets.  if it does exist then add the genes
@@ -102,7 +102,7 @@ public void run(TaskMonitor taskMonitor) throws IOException {
 
 			//finished parsing that geneset
 			//add the current geneset to the hashmap of genesets
-			GeneSet gs = new GeneSet(name, description, builder.build());
+			GeneSet gs = GeneSet.createDavid(name, description, builder.build(), category);
 			genesets.put(name, gs);
 
 			//The 5th column is the nominal p-value
@@ -148,8 +148,11 @@ public void run(TaskMonitor taskMonitor) throws IOException {
 			}
 
 		}
-		if(FDR)
+		
+		dataset.getMap().getParams().setDavid(true);
+		if(FDR) {
 			dataset.getMap().getParams().setFDR(FDR);
+		}
 	}
 
 }

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/style/EMStyleBuilder.java

@@ -74,6 +74,7 @@ public static class Columns {
 		// Common attributes that apply to the entire network
 		public static final ColumnDescriptor<String> NODE_NAME = new ColumnDescriptor<>("Name", String.class);
 		public static final ColumnDescriptor<String> NODE_GS_DESCR = new ColumnDescriptor<>("GS_DESCR", String.class);
+		public static final ColumnDescriptor<String> NODE_DAVID_CATEGORY = new ColumnDescriptor<>("david_category", String.class);
 		public static final ColumnDescriptor<String> NODE_DATASOURCE = new ColumnDescriptor<>("GS_datasource", String.class);
 		public static final ColumnDescriptor<String> NODE_DATASOURCEID = new ColumnDescriptor<>("GS_datasource_id", String.class);
 		public static final ColumnDescriptor<String> NODE_GS_TYPE  = new ColumnDescriptor<>("GS_Type", String.class);

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/task/CreateEMNetworkTask.java

@@ -137,6 +137,11 @@ private Map<String, CyNode> createNodes(CyNetwork network) {
 						Columns.NODE_DATASOURCEID.set(row, prefix, null, geneSet.getDatasourceId().get());
 					}
 				}
+				if(map.getParams().isDavid()) {
+					if(geneSet.getDavidCategory().isPresent()) {
+						Columns.NODE_DAVID_CATEGORY.set(row, prefix, null, geneSet.getDavidCategory().get());
+					}
+				}
 			}
 
 			Columns.NODE_GS_TYPE.set(row, prefix, null, Columns.NODE_GS_TYPE_ENRICHMENT);
@@ -223,6 +228,9 @@ private CyTable createNodeColumns(CyNetwork network) {
 			Columns.NODE_DATASOURCE.createColumn(table, prefix, null);
 			Columns.NODE_DATASOURCEID.createColumn(table, prefix, null);
 		}
+		if(params.isDavid()) {
+			Columns.NODE_DAVID_CATEGORY.createColumn(table, prefix, null);
+		}
 
 		for (EMDataSet dataset : map.getDataSetList()) {
 			Columns.NODE_PVALUE.createColumn(table, prefix, dataset);