Skip to content

Latest commit

 

History

History
74 lines (57 loc) · 2.87 KB

README.md

File metadata and controls

74 lines (57 loc) · 2.87 KB
Raw

met4j-graph

Met4j module for graph-based topological analysis of metabolic networks

Installation

cd met4j-graph;
mvn clean install;

Using met4j-graph in a maven project

Put this in the pom.xml file of your project:

<dependencies>
...
		<dependency>
			<groupId>fr.inrae.toulouse.metexplore</groupId>
			<artifactId>met4j-graph</artifactId>
			<version>0.0.1-SNAPSHOT</version>
		</dependency>
...
</dependencies>

The version number will vary in the future.

Documentation

Examples can be found in https://forgemia.inra.fr/metexplore/tutorialmet4j in the package fr.inrae.toulouse.metexplore.tutorialmet4j.met4j_graph.

Creating a met4j graph

From a BioNetwork build the met4J core API, several graphs can be created: Compound graph, where nodes represent metabolites, and directed edges links two compounds if they are respectively substrates and products of the same reaction. A reaction is thus represented by a set of edges. Reaction graph, where nodes represent reactions, and directed edges links two reactions if one produces a substrate of the other. Bipartite graph, where nodes can represent both metabolites and reactions. Directed edges links substrate metabolites to their consuming reactions, and reactions to their products metabolites.

//Read BioNetwork
JsbmlReader reader = new JsbmlReader(pathOfYourSbmlFile);
BioNetwork network = reader.read();

//Build graphs
Bionetwork2BioGraph builder = new Bionetwork2BioGraph(network);
CompoundGraph cgraph = builder.getCompoundGraph();
ReactionGraph rgraph = builder.getReactionGraph();
BipartiteGraph bgraph = builder.getBipartiteGraph();
Performing network anlysis using met4j-graph

All graphs in met4j-graph extend BioGraph, a wrapper class for JGraphT DirectedWeightedMultigraph, and add convenience methods relevant in metabolic network analysis.

Metabolic graphs can be analyzed using the JGraphT generic algorithms or met4J-graph implementation which include more domain-specific analysis, such as load points, choke points, precursor network or scope network.

met4J-graph also includes many weighting functions tailored for metabolic network analysis, including Chemical similarity or degree squared as well as edge weighting for Markov-Chain analysis.

//perform chemical similarity weighting
WeightingPolicy wp = new SimilarityWeightPolicy(FingerprintBuilder.KLEKOTAROTH);
wp.setWeight(cgraph);
//Compute lightest path
ShortestPath pathSearch = new ShortestPath<BioMetabolite, ReactionEdge, CompoundGraph>(cgraph);
BioPath<BioMetabolite, ReactionEdge> path = 
        pathSearch.getShortest(cgraph.getVertex("glucose"), cgraph.getVertex("pyruvate"));
return path.getLength();

Finally, met4J-graph provides export to gml and xgmml files to perform analysis and visualization in Cytoscape.