Vincent Danos, Chair in Computational Systems Biology, School of Informatics, University of Edinburgh

Title: Esse Est Percipi (To be is to be Perceived) Endogenous Reduction Techniques for Rule-Based Models of Signaling Pathways [Joint work with J. Feret, J. Krivine,R. Harmer, and W. Fontana]

A rule-based model consists of rules stipulating under which conditions (and at which rate) specific protein-protein interactions can occur. When such conditions are local, rule-based models cope well with the combinatorial explosion generated by post-translational modifications and complex formation. The exploration of the implied dynamics, however, requires a stochastic simulation, which can become very costly. We describe a generic lumping technique using coarse- grained variables that are determined by what the dynamics of the original system can distinguish, and thus can be considered as endogenous information carriers. The method rests entirely on static analysis, is exact, does not depend on the rate constant of the rules of interest, and obtains spectacular reduction in the dimension of actual pathway models.