Hybrid moment computation algorithm for biochemical reaction networks

Abstract

Moment computation is essential to the analysis of stochastic kinetic models of biochemical reaction networks. It is often the case that the moment evolution, usually the first and the second moment evolutions over time, is all the information of interest. However, potential approaches to moment computation, specifically, the moment closure method and the exact stochastic simulation method, have their significant deficiency. The former, despite its computational efficiency, is essentially an approximation to the real solution and thus is lack of inaccuracy at certain conditions, while the computational inefficiency makes the usage of the latter limited to the networks with small number of molecules. A hybrid moment computation algorithm is therefore proposed by integrating the moment closure method and the exact stochastic simulation algorithms. The moment closure method and the stochastic simulation algorithm operate by turns to achieve an optimal balance between the efficiency due to the moment closure method and the accuracy due to the stochastic simulation. The hybrid algorithm is applied to a Dictyostelium cAMP oscillation network. The simulation results illustrate the effectiveness of the algorithm.