Convergence dynamics of biochemical pathway steady state stochastic global optimization

The stochastic nature of convergence of steady state stochastic global optimization methods in design optimization tasks with steady state precondition is a hardly predictable step in development of industrially efficient strains of microorganisms. The properties of convergence dynamics of evolutionary programming (EP) and particle swarm (PS) depending on complexity of kinetic equations within the same model are studied optimizing yeast glycolysis for ethanol production by COPASI software adjusting parameters of three combinations of five reactions out of fifteen enzymatic reactions. Results indicate significant differences in the convergence dynamics between different combinations. 50-fold difference in convergence time as well as possible stagnation at local optima was observed. The choice of optimization method and duration of optimization runs should be based on number of tests on the convergence quality, speed and repeatability.