Convergence dynamics of biochemical models to the global optimum

Stochastic nature of convergence of steady state stochastic global optimization methods causes several seemingly attractive approaches to reduce the length of the optimization procedure. The properties of convergence dynamics of evolutionary programming (EP) and particle swarm (PS) are studied optimizing yeast glycolysis by COPASI software adjusting parameters of one, five, ten and fifteen reactions with five identical runs for each case. Results indicate the potential and risks of shortening the optimization time improving the possibilities of systematic search of adjustable parameter combinations. The choice of optimization method depending on the model size and the number of adjustable parameters should be based on number of tests on the convergence quality, speed and repeatability.