Chapter

Evolutionary Computation, Machine Learning and Data Mining in Bioinformatics

Volume 5483 of the series Lecture Notes in Computer Science pp 116-127

A Comparison of Genetic Algorithms and Particle Swarm Optimization for Parameter Estimation in Stochastic Biochemical Systems

  • Daniela BesozziAffiliated withDipartimento di Informatica e Comunicazione, Università di Milano
  • , Paolo CazzanigaAffiliated withDipartimento di Informatica, Sistemistica e Comunicazione, Università di Milano-Bicocca
  • , Giancarlo MauriAffiliated withDipartimento di Informatica, Sistemistica e Comunicazione, Università di Milano-Bicocca
  • , Dario PesciniAffiliated withDipartimento di Informatica, Sistemistica e Comunicazione, Università di Milano-Bicocca
  • , Leonardo VanneschiAffiliated withDipartimento di Informatica, Sistemistica e Comunicazione, Università di Milano-Bicocca

Abstract

The modelling of biochemical systems requires the knowledge of several quantitative parameters (e.g. reaction rates) which are often hard to measure in laboratory experiments. Furthermore, when the system involves small numbers of molecules, the modelling approach should also take into account the effects of randomness on the system dynamics. In this paper, we tackle the problem of estimating the unknown parameters of stochastic biochemical systems by means of two optimization heuristics, genetic algorithms and particle swarm optimization. Their performances are tested and compared on two basic kinetics schemes: the Michaelis-Menten equation and the Brussellator. The experimental results suggest that particle swarm optimization is a suitable method for this problem. The set of parameters estimated by particle swarm optimization allows us to reliably reconstruct the dynamics of the Michaelis-Menten system and of the Brussellator in the oscillating regime.