Analysis of a genetic programming algorithm for association studies

Abstract

In this paper a Genetic Programming algorithm for genetic association studies is reconsidered. It is shown, that the application field of the algorithm is not restricted to genetic association studies, but that the algorithm can also be applied to logic minimization problems. In the context of multi-valued logic minimization on incompletely specified truth tables it outperforms existing algorithms. In addition, the facilities of the algorithm in the original application field are complemented by new results and experiments. This includes answers to the open questions of how to automatically choose the best individual in the last population and whether crossover is necessary for the algorithm.