Multi-objective steady state optimization of biochemical reaction networks using a constrained genetic algorithm

A hybrid genetic algorithm-based method to solve constrained multi-objective optimization problems is proposed. Considering operation around a steady state of a dynamical system, the task of the algorithm consists on finding a set of optimal, but constrained solutions. The method is exemplified on a (bio)chemical reaction network in Saccharomyces cerevisiae. In the steady state the model reduces to a system of non-linear equations which must be solved by a search method. This iterative search was integrated into a genetic algorithm in order to look up for optimal steady states. The basic idea is to use individuals of the genetic algorithm as starting points for the search algorithm. The optimization goal was to simultaneously maximize ethanol production and reduce metabolic burden. Two alternative kinetic approaches are compared to Michaelis Menten-type kinetics: a S-System and a generalized mass action model, both based on Power-Law kinetics.