MODELING AND OPTIMAL CONTROL OF ESCHERICHIA COLI GENETIC CIRCUITS WITH INTRINSIC STOCHASTICITY

Abstract

This work considers aspects of the pathway performance optimization of <em>Escherichia coli</em> genetic circuits. A characteristic of such circuits is that some molecules are present in very low quantities. This leads to the use of a simulation environment, Stochastic Petri Nets (SPNs), which is appropriate for obtaining stochastic uncertainty information; however the direct use of such models in optimization poses significant computational problems. We propose to approximate SPNs by Langevin-type models that incorporate stochastic variance information generated by SPN simulations, but which are simpler and potentially applicable for direct use in optimization. We discuss such models in the context of σ<sup>32</sup>-mediated stress responses. Simulation results for both ethanol stress response and recombinant protein production match well with experimental data from the literature. The optimal control problem that maximizes the production of active protein through antisense RNA induction is then studied.