Automatic Parameterisation of Stochastic Petri Net Models of Biological Networks

Abstract

Stochastic simulations are able to capture the fine grain behaviour and randomness of outcome of biological networks not captured by deterministic techniques. As such they are becoming an increasingly important tool in the biological community. However, current efforts in the stochastic simulation of biological networks are hampered by two main problems: firstly the lack of complete knowledge of kinetic parameters; and secondly the computational cost of the simulations. In this paper we investigate these problems using the framework of stochastic Petri nets. We present a new stochastic Petri net simulation tool NASTY which allows large numbers of stochastic simulations to be carried out in parallel. We then begin to address the important problem of incomplete knowledge of kinetic parameters by developing a distributed genetic algorithm, based on NASTY's simulation engine, to parameterise stochastic networks. Our algorithm is able to successfully estimate kinetic parameters to replicate a system's behaviour and we illustrate this by presenting a case study in which the kinetic parameters are derived for a stochastic model of the stress response pathway in the bacterium E.coli.