Filter signature edges by dataset.

Refs #462

Author: mikekucera

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

@@ -18,6 +18,7 @@
 import java.util.stream.Collectors;
 
 import org.baderlab.csplugins.enrichmentmap.model.EMDataSet.Method;
+import org.baderlab.csplugins.enrichmentmap.style.EMStyleBuilder;
 import org.baderlab.csplugins.enrichmentmap.util.NetworkUtil;
 import org.cytoscape.model.CyNetwork;
 import org.cytoscape.model.CyNetworkManager;
@@ -420,16 +421,74 @@ public static Set<Long> getNodesIntersection(Collection<? extends AbstractDataSe
 		return getIntersection(dataSets, AbstractDataSet::getNodeSuids);
 	}
 
+	
 	/**
 	 * Returns the SUIDs for all the gene-sets in the given collection of DataSets.
 	 * Each returned gene-set is contained in at least one of the given DataSets.
 	 * 
 	 * Note, this will only return distinct edges, not compound edges.
+	 * 
+	 * This method considers edges that are connected to signature gene sets to be part of the
+	 * signature data set.
 	 */
 	public static Set<Long> getEdgesUnion(Collection<? extends AbstractDataSet> dataSets) {
 		return getUnion(dataSets, AbstractDataSet::getEdgeSuids);
 	}
 
+	/**
+	 * This method checks if signature edges are actually part of one of the given regular data sets
+	 * when including them or not.
+	 */
+	public static Set<Long> getEdgesUnionForFiltering(Collection<? extends AbstractDataSet> dataSets, EnrichmentMap map, CyNetwork network) {
+		List<EMDataSet> regularDatasets = new ArrayList<>();
+		List<EMSignatureDataSet> signatureDatasets = new ArrayList<>();
+		
+		for(var ds : dataSets) {
+			if(ds instanceof EMDataSet) {
+				regularDatasets.add((EMDataSet)ds);
+			} else if(ds instanceof EMSignatureDataSet) {
+				signatureDatasets.add((EMSignatureDataSet)ds);
+			}
+		}
+		
+		Set<Long> regularEdgesUnion = getUnion(regularDatasets, AbstractDataSet::getEdgeSuids);
+		
+		if(map == null || network == null || signatureDatasets.isEmpty()) {
+			return regularEdgesUnion;
+		}
+		
+		// The problem with Signature edges is that they are associated with the Signature Data Set,
+		// not the data set they were computed against. This information is only available in the
+		// edge table. So we have to go through all the Signature edges and check if they are associated
+		// with any of the regular data sets.
+		
+		Set<Long> allEdges = new HashSet<>();
+		allEdges.addAll(regularEdgesUnion);
+		
+		
+		Set<String> regularDataSetNames = new HashSet<>();
+		for(var ds : regularDatasets) {
+			regularDataSetNames.add(ds.getName());
+		}
+		
+		var edgeTable = network.getDefaultEdgeTable();
+		
+		for(var sigDS : signatureDatasets) {
+			for(Long suid : sigDS.getEdgeSuids()) {
+				if(edgeTable.rowExists(suid)) {
+					var row = edgeTable.getRow(suid);
+					String prefix = map.getParams().getAttributePrefix();
+					String dsname = EMStyleBuilder.Columns.EDGE_DATASET.get(row, prefix);
+					if(regularDataSetNames.contains(dsname)) {
+						allEdges.add(suid);
+					}
+				}
+			}
+		}
+		
+		return allEdges;
+	}
+	
 	/**
 	 * Returns the SUIDs for all the gene-sets in the given collection of DataSets.
 	 * Each returned gene-set is contained all of the given DataSets.

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

@@ -30,19 +30,29 @@
 
 public class SelectNodesEdgesTask extends AbstractTask {
 
+	private final EnrichmentMap map;
 	private final CyNetworkView networkView;
 	private final Set<AbstractDataSet> dataSets;
 	private final boolean distinctEdges;
 
 	@Inject private CyEventHelper eventHelper;
 
 	public interface Factory {
-		SelectNodesEdgesTask create(CyNetworkView networkView, Set<AbstractDataSet> dataSets, boolean distinctEdges);
+		SelectNodesEdgesTask create(
+			EnrichmentMap map,
+			CyNetworkView networkView, 
+			Set<AbstractDataSet> dataSets,
+			boolean distinctEdges);
 	}
 
 	@Inject
-	public SelectNodesEdgesTask(@Assisted CyNetworkView networkView, @Assisted Set<AbstractDataSet> dataSets,
-			@Assisted boolean distinctEdges) {
+	public SelectNodesEdgesTask(
+			@Assisted EnrichmentMap map,
+			@Assisted CyNetworkView networkView, 
+			@Assisted Set<AbstractDataSet> dataSets,
+			@Assisted boolean distinctEdges
+	) {
+		this.map = map;
 		this.networkView = networkView;
 		this.dataSets = dataSets;
 		this.distinctEdges = distinctEdges;
@@ -58,7 +68,7 @@ public void run(TaskMonitor tm) {
 		final Set<Long> dataSetEdges;
 
 		if (distinctEdges) {
-			dataSetEdges = EnrichmentMap.getEdgesUnion(dataSets);
+			dataSetEdges = EnrichmentMap.getEdgesUnionForFiltering(dataSets, map, networkView.getModel());
 		} else {
 			dataSetEdges = new HashSet<>();
 

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/task/postanalysis/CreatePANetworkTask.java

@@ -329,7 +329,7 @@ private void createEdge(SimilarityKey similarityKey, CyNetwork network, CyNetwor
 		Columns.EDGE_OVERLAP_GENES.set(row, prefix, null, geneList);
 		Columns.EDGE_OVERLAP_SIZE.set(row, prefix, null, genesetSimilarity.getSizeOfOverlap());
 		Columns.EDGE_SIMILARITY_COEFF.set(row, prefix, null, genesetSimilarity.getSimilarityCoeffecient());
-		Columns.EDGE_DATASET.set(row, prefix, null, similarityKey.getName());
+		Columns.EDGE_DATASET.set(row, prefix, null, similarityKey.getName()); // the similarityKey name is the same as the dataset name
 		Columns.EDGE_SIG_DATASET.set(row, prefix, null, sigDataSet.getName());
 
 		String dataset = genesetSimilarity.getDataSetName();

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/task/postanalysis/PASimilarityTaskParallel.java

@@ -51,14 +51,9 @@ public class PASimilarityTaskParallel extends CancellableParallelTask<Map<Simila
 
 
 	public static interface Factory {
-		PASimilarityTaskParallel create(PostAnalysisParameters params, EnrichmentMap map);
 		PASimilarityTaskParallel create(PostAnalysisParameters params, EnrichmentMap map, List<EMDataSet> dataSets);
 	}
 
-	@AssistedInject
-	public PASimilarityTaskParallel(@Assisted PostAnalysisParameters params, @Assisted EnrichmentMap map) {
-		this(params, map, map.getDataSetList());
-	}
 
 	@AssistedInject
 	public PASimilarityTaskParallel(@Assisted PostAnalysisParameters params, @Assisted EnrichmentMap map, @Assisted List<EMDataSet> dataSets) {
@@ -138,10 +133,11 @@ public Map<SimilarityKey,SignatureGenesetSimilarity> compute(TaskMonitor tm, Exe
 								boolean passesCutoff = metric.passes(value);
 								comparison.setPassesCutoff(passesCutoff);
 
+								// Very important that the SimilarityKey name is the dataset name.
+								// This gets picked up by the CreatePANetworkTask and set on the "Dataset" edge column.
 								SimilarityKey key = new SimilarityKey(hubName, geneSetName, INTERACTION, dataSet.getName());
 								geneSetSimilarities.put(key, comparison);
-							}
-							
+							} 
 						}
 					}
 				}

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/task/postanalysis/RemoveSignatureDataSetsTask.java

@@ -67,6 +67,7 @@ private Set<Long> getNodesToDelete() {
 	}
 
 	private Set<Long> getEdgesToDelete() {
+		// TODO Won't deleting the nodes automatically delete the adjacent edges???
 		Set<Long> edgesToDelete = EnrichmentMap.getEdgesUnion(signatureDataSets);
 		return edgesToDelete;
 	}

EnrichmentMapPlugin/src/main/java/org/baderlab/csplugins/enrichmentmap/view/control/ControlPanelMediator.java

@@ -572,6 +572,7 @@ private void addListeners(EMViewControlPanel viewPanel, EnrichmentMap map) {
 
 		viewPanel.getDataSetSelector().getSelectNodesMenuItem().addActionListener(evt -> {
 			selectNodesEdges(
+				map,
 				viewPanel.getNetworkView(),
 				viewPanel.getDataSetSelector().getSelectedItems(),
 				map.getParams().getCreateDistinctEdges()
@@ -1050,8 +1051,8 @@ private void showContextMenu(final JPopupMenu contextMenu, final MouseEvent e) {
 		});
 	}
 
-	private void selectNodesEdges(CyNetworkView netView, Set<AbstractDataSet> dataSets, boolean distinctEdges) {
-		SelectNodesEdgesTask task = selectNodesEdgesTaskFactory.create(netView, dataSets, distinctEdges);
+	private void selectNodesEdges(EnrichmentMap map, CyNetworkView netView, Set<AbstractDataSet> dataSets, boolean distinctEdges) {
+		SelectNodesEdgesTask task = selectNodesEdgesTaskFactory.create(map, netView, dataSets, distinctEdges);
 		dialogTaskManager.execute(new TaskIterator(task));
 	}
 
@@ -1079,21 +1080,19 @@ public void actionPerformed(ActionEvent evt) {
 
 			// Find nodes and edges that must be displayed
 			Set<CyNode> filteredInNodes = getFilteredInNodes(selectedDataSets);
-			
 			if (cancelled)
 				return;
 
 			Set<CyEdge> filteredInEdges = getFilteredInEdges(selectedDataSets);
-			
 			if (cancelled)
 				return;
 
 			// Run Task
 			task = filterNodesEdgesTaskFactory.create(map, netView, filteredInNodes, filteredInEdges, filterMode);
 			dialogTaskManager.execute(new TaskIterator(task), TaskUtil.allFinished(finishStatus -> {
-					task = null;
-					if (!cancelled)
-						netView.updateView();
+				task = null;
+				if (!cancelled)
+					netView.updateView();
 			}));
 		}
 
@@ -1138,20 +1137,23 @@ private Set<CyEdge> getFilteredInEdges(Set<AbstractDataSet> selectedDataSets) {
 			EMCreationParameters params = map.getParams();
 
 			// Compound edges are not associated with a specific data set
-			Set<Long> dataSetEdges = EnrichmentMap.getEdgesUnion(selectedDataSets);
+			Set<Long> dataSetEdges = EnrichmentMap.getEdgesUnionForFiltering(selectedDataSets, map, netView.getModel());
 
-			if (viewPanel.getSimilaritySliderPanel() != null)
-				return getFilteredInEdges(viewPanel.getSimilaritySliderPanel(), map, netView,
-						params.getSimilarityCutoffColumnNames(), dataSetEdges);
+			var sliderPanel = viewPanel.getSimilaritySliderPanel();
+			if (sliderPanel != null) {
+				var names = params.getSimilarityCutoffColumnNames();
+				return getFilteredInEdges(sliderPanel, map, netView, names, dataSetEdges);
+			}
 
 			Set<CyEdge> filteredInEdges = new HashSet<>();
 			CyNetwork net = netView.getModel();
 			boolean distinct = params.getCreateDistinctEdges();
 
 			if (distinct) {
 				for (CyEdge e : net.getEdgeList()) {
-					if (dataSetEdges.contains(e.getSUID()))
+					if (dataSetEdges.contains(e.getSUID())) {
 						filteredInEdges.add(e);
+					}
 				}
 			} else {
 				// If compound edges, all edges are filtered in, no matter the selected data sets