Identifying coherent subnetworks in genome scale metabolic networks

Constraint-based methods to analyse metabolic networks require the classification of metabolites into external compounds that are exchanged with the system environment, and internal ones that form intermediate steps on the metabolic pathways and which are therefore subject to stoichiometry constraints in the steady state. For eukaryotic cells it is not practical to analyse the complete cellular network, and so the subnetwork that describes the specific phenomena under study needs to be extracted. This study focuses on flavonoid production in the model plant Arabidopsis Thaliana. Software was developed to extract the full Arabidopsis metabolic network from the AraCyc database, in a format required by standard network analysis packages. Considerable processing is also required to reconcile information about reaction directions from different database tables and to classify metabolites as internal or external to the subnetwork. The paper outlines the strategies implemented in the software to address these issues