A Comparative Study of Genetic Network Modeling Using Predator-Prey System

Genetic network modeling is an inverse problem in the sense that given limited amount of experimental data of gene expressions, a dynamic model is sought to fit the data for inferences of biological processes. In this study, a well-known ecological system, the predator-prey differential system, is used to model the dynamics. A fourth order Runge-Kutta algorithm is employed to solve a predator-prey system, given coefficients of the system. Deviations between numerically computed solutions and the given data are used to assess the fitness of these coefficients. Different population based search algorithms including particle swarm optimization, differential evolution and genetic algorithm are used to find appropriate network coefficients. Synthetic data and real data from E. Coli SOS DNA repair process were used to verify performance of the proposed methods. It was found that genetic algorithm has provided the best solution among the three algorithms considered in this study.