Maximum likelihood methods in biology revisited with tools of computational intelligence

We investigate the problem of identification of genes correlated with the occurrence of diseases in a given population. The classical method of parametric linkage analysis is combined with newer tools and results are achieved on a model problem. This traditional method has advantages over non-parametric methods, but these advantages have been difficult to realize due to their high computational cost. We study a class of Evolutionary Algorithms from the Computational Intelligence literature which are designed to cut such costs considerably for optimization problems. We outline the details of this algorithm, called Particle Swarm Optimization, and present all the equations and parameter values we used to accomplish our optimization. We view this study as a launching point for a wider investigation into the leveraging of computational intelligence tools in the study of complex biological systems.