Item

Abstract

A rigorous, detailed description of signaling networks gives rise to huge models, while an a priori simplified model relies on assumptions difficult to prove. Therefore, models which are manageable, yet retain the essential properties of the real network, are desirable. Herein, we first describe a method that addresses the combinatorial explosion of the number of states due to the ability of proteins to bind multiple partners via different domains. We show how a linear state transformation together with the application of the system-theoretical concept of observability allow a dramatic reduction of the number of states to be considered. Secondly, we present an approach for a further reduction of the models, based on a decomposition of the model into modules. The resulting subunits are analyzed via simulation studies, leading to the identification of less complex non-linear models showing approximately the same input/output behavior, which can replace the complex modules in the whole model.