Extending the Metabolic Network of Ectocarpus Siliculosus using Answer Set Programming
Résumé
Metabolic network reconstruction is of great biological relevance be- cause it offers a way to investigate the metabolic behavior of organisms. However, such a reconstruction remains a difficult task at both the biological and compu- tational level. Building on previous work establishing an ASP-based approach to this problem, we present a report from the field resulting in the discovery of new biological knowledge. In fact, for the first time ever, we automatically reconstructed a metabolic network for a macroalgea. We accomplished this by taking advantage of ASP's integrated optimization and enumeration capacities. Both tasks have been modeled in an improved ASP problem representation, in- corporating the concept of reversible reactions. Interestingly, it turned out that optimization highly benefits from the usage of unsatisfiable cores available in the ASP solver unclasp. Finally, applied to Ectocarpus siliculosus, only the combi- nation of unclasp and clasp allowed us to obtain a metabolic network able to produce all recoverable metabolites among the experimentally measured ones. Moreover, 70% of the identified reactions are supported by the existence of an homologous enzyme in Ectocarpus siliculosus, confirming the quality of the re- constructed network from a biological point of view.