A team of genetic algorithms for the multiple sequence alignment problem: preliminary results

Simultaneous alignment of biological sequences is a common task within the bioinformatics area, since it settles the bases for later analyzes of protein families, like the modeling of similarities, phylogeny reconstruction, or to show the conserved and variable sites within a protein family. The multiple alignment of sequences is a complex combinatorial optimization problem, in fact it belongs to the NP-hard class when the number of sequences is more than two. For this kind of problems evolutionary algorithms have achieved good approximate solutions. On the other hand, there is a methodology known as team algorithms, i.e., a set of algorithms, which can solve a problem when they work together, although incapable of doing so individually. This work consists of the design of a genetic algorithm capable of finding a global near optimal multiple alignment for the case of small alignments, beginning from completely disaligned sequences and applying concepts of team algorithms, where the team is conformed by genetic algorithms with different objective functions. The obtained results show a significant improvement in the quality of the alignments generated by the team genetic algorithm in comparison with the results found by the simple genetic algorithm. This happens when a high correlation between the objective functions that conforms the team exists.