GraphiteImport.java

package edu.uconn.tripoint.pathwayimport;

import java.util.Map;
import java.util.TreeMap;

import org.rosuda.REngine.REXPMismatchException;
import org.rosuda.REngine.REngineException;
import org.rosuda.REngine.Rserve.RConnection;
import org.rosuda.REngine.Rserve.RserveException;

import edu.uconn.tripoint.pathway.Gene;
import edu.uconn.tripoint.pathway.Pathway;
import edu.uconn.tripoint.pathway.PathwayEdge;

/**
 * Imports pathways from the Graphite R package using RServe.
 *
 */
public class GraphiteImport {
	
	//All pathways merged into one.
	private Pathway _globalpathway;
	//Each individual pathway for each pathway in the library
	private Pathway[] _pathways; 
	//All genes in the pathways.  Used to update values after import.
	private Map<String, Gene> _genes; 
	
	/**
	 * Imports a set of pathways from the Graphite R package.
	 * Requires RServe to be running in R.
	 * @param pathwaylibrary	
	 * The name of the graphite pathway library 
	 * to use ("kegg", "reactome", "biocarta", "humancyc", "nci", "panther")
	 */
	public GraphiteImport(String pathwaylibrary, String gid){
		Map<String, PathwayEdge> edgetypemapping = getEdgeTypeMapping();
		startRServe();
		RConnection c;
		System.out.println("Starting import.");
		try {
			//Creates an R Connection and loads the pathways in R.
			c = new RConnection();
			c.parseAndEval("library(graphite)");
			c.parseAndEval("plib <- pathways(\"hsapiens\", \""+pathwaylibrary+"\")");
			c.parseAndEval("cpathwaylist <- convertIdentifiers(plib, \""+gid+"\")");
			int numpathways = c.parseAndEval("length(cpathwaylist)").asInteger();

			//Set the global variables
			_pathways = new Pathway[numpathways];
			_globalpathway = new Pathway("All "+pathwaylibrary);
			_genes = new TreeMap<String, Gene>();
			
			System.out.println("Creating graphs.");
			//Import each pathway
			for(int i = 0; i < numpathways; i++){
				int rindex = (i+1);
				String pathwayname = c.parseAndEval("names(cpathwaylist)["+rindex+"]").asString();
				c.parseAndEval("curpath = cpathwaylist[["+rindex+"]]");
				c.parseAndEval("ed <- edges(curpath)");

				String[] sources = c.parseAndEval("ed$src").asStrings();
				String[] destinations = c.parseAndEval("ed$dest").asStrings();
				String[] edgetype = c.parseAndEval("ed$type").asStrings();

				System.out.println("Loading "+pathwayname);
				Pathway p = new Pathway(pathwaylibrary+":"+pathwayname);
				_pathways[i] = p;
				
				//Add genes/nodes to the graph/network
				for(int j = 0; j < sources.length; j++){
					if(!_genes.containsKey(sources[j])){
						_genes.put(sources[j], new Gene(sources[j]));
					}
					Gene g = _genes.get(sources[j]);
					p.addGene(g);
					_globalpathway.addGene(g);
				}
				for(int j = 0; j < destinations.length; j++){
					if(!_genes.containsKey(destinations[j])){
						_genes.put(destinations[j], new Gene(destinations[j]));
					}
					Gene g = _genes.get(destinations[j]);
					p.addGene(g);
					_globalpathway.addGene(g);
				}
				
				//Add the edges
				for(int j = 0; j < edgetype.length; j++){
					PathwayEdge pe = edgetypemapping.get(edgetype[j]);
					if(pe != null){
						Gene g1 = _genes.get(sources[j]);
						Gene g2 = _genes.get(destinations[j]);
						try {
							p.addEdge(g1, g2, pe);
							_globalpathway.addEdge(g1,g2,pe);
						} catch (Exception e) {
							e.printStackTrace();
						}
					}
				}
			}
			c.close();
			
		} catch (RserveException e) {
			e.printStackTrace();
		} catch (REXPMismatchException e) {
			e.printStackTrace();
		} catch (REngineException e) {
			e.printStackTrace();
		}

	}
	
	//Contains the mapping of edge types from Graphite to Activation, Inhibition, and Association
	private Map<String, PathwayEdge> getEdgeTypeMapping(){
		Map<String, PathwayEdge> rv = new TreeMap<String, PathwayEdge>();
		rv.put("Binding", PathwayEdge.ASSOCIATION());
		rv.put("Control(In: ACTIVATION of BiochemicalReaction)", PathwayEdge.ACTIVATION());
		rv.put("Control(In: ACTIVATION of ComplexAssembly)", PathwayEdge.ACTIVATION());
		rv.put("Control(In: ACTIVATION of Degradation)", PathwayEdge.ACTIVATION());
		rv.put("Control(In: ACTIVATION of Transport)", PathwayEdge.ACTIVATION());
		rv.put("Control(indirect)", null);
		rv.put("Control(In: INHIBITION-COMPETITIVE of BiochemicalReaction)", PathwayEdge.INHIBITION());
		rv.put("Control(In: INHIBITION of BiochemicalReaction)", PathwayEdge.INHIBITION());
		rv.put("Control(In: INHIBITION of ComplexAssembly)", PathwayEdge.INHIBITION());
		rv.put("Control(In: INHIBITION of Transport)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: ACTIVATION of ACTIVATION)", PathwayEdge.ACTIVATION());
		rv.put("Control(Out: ACTIVATION of BiochemicalReaction)", PathwayEdge.ACTIVATION());
		rv.put("Control(Out: ACTIVATION of ComplexAssembly)", PathwayEdge.ACTIVATION());
		rv.put("Control(Out: ACTIVATION of TemplateReaction)", PathwayEdge.ACTIVATION());
		rv.put("Control(Out: ACTIVATION of Transport)", PathwayEdge.ACTIVATION());
		rv.put("Control(Out: INHIBITION-COMPETITIVE of BiochemicalReaction)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: INHIBITION of ACTIVATION)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: INHIBITION of BiochemicalReaction)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: INHIBITION of ComplexAssembly)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: INHIBITION of TemplateReaction)", PathwayEdge.INHIBITION());
		rv.put("Control(Out: INHIBITION of Transport)", PathwayEdge.INHIBITION());
        rv.put("Process", PathwayEdge.ACTIVATION());
        rv.put("Process(activation)", PathwayEdge.ACTIVATION());
        rv.put("Process(binding/association)", PathwayEdge.ASSOCIATION());
        rv.put("Process(BiochemicalReaction)", PathwayEdge.ACTIVATION());
        rv.put("Process(demethylation)",  PathwayEdge.ASSOCIATION());
        rv.put("Process(dephosphorylation)",  PathwayEdge.ASSOCIATION());
        rv.put("Process(dissociation)", null);
        rv.put("Process(expression)", PathwayEdge.ACTIVATION()); //Setting this to activation
        rv.put("Process(glycosylation)", PathwayEdge.ASSOCIATION());
        rv.put("Process(indirect)", null);
        rv.put("Process(indirect effect)", null);
        rv.put("Process(inhibition)", PathwayEdge.INHIBITION());
        rv.put("Process(methylation)", PathwayEdge.ASSOCIATION());
        rv.put("Process(missing)", null);
        rv.put("Process(missing interaction)", null);
        rv.put("Process(phosphorylation)",  PathwayEdge.ASSOCIATION());
        rv.put("Process(repression)", PathwayEdge.INHIBITION());
        rv.put("Process(state change)",  PathwayEdge.ASSOCIATION());
		return rv;
	}
	
	public Pathway getGlobalPathway(){
		return _globalpathway;
	}
	
	public Pathway[] getPathways(){
		return _pathways;
	}
	
	public Map<String, Gene> getGenes(){
		return _genes;
	}
	
	private void startRServe(){
		//TODO start RServe within this application
	}
	
}