Evolving Consensus Sequences with a Genetic Algorithm

In this paper we present an approach that employs a genetic algorithm (GA) to evolve the consensus sequences for DNA and protein sequences alignments. We have developed an encoding scheme such that the number of generations needed to find the optimal solution remains approximately the same regardless the number of sequences. The complexity instead depends only on the length of the consensus sequence and similarities among sequences. The objective function gives the sum-of-pairs (SP) scores that serve as the fitness values. We have devised a residue profiling technique that further simplifies the calculations of the SP scores. Furthermore, to facilitate the quantitative studies of our GA approach, we have developed a simulation program that incorporates the most commonly used evolutionary models and generates biologically sound sequences. We performed several experiments and compared the results with the most commonly used heuristic alignment program Clustal W [18, 19]. We conclude our research with detailed analysis and demonstrate that our GA approach offers an attractive and competitive alternative to the heuristic approach.