Performance analysis of Multiobjective Artificial Bee Colony implementations for phylogenetic reconstruction

The inference of phylogenetic relationships represents one of the most challenging problems in bioinformatics. The increasing availability of biological data motivates the development of new algorithmic designs to conduct phylogenetic analyses on exponentially increasing search spaces. Bioinspired metaheuristics have arisen as a useful approach to address this problem, introducing different search strategies according to the way phylogenetic trees are represented and handled by the algorithm. In this work, we study the multiobjective and biological performance achieved by different Multiobjective Artificial Bee Colony implementations based on direct (tree-based) and indirect (distance-based) individual representations. Experiments on four real nucleotide data sets show meaningful differences in multiobjective performance between the analyzed approaches, obtaining significant biological results in comparison with other state-of-the-art phylogenetic methods.