Phylogeny inference using a multi-objective evolutionary algorithm with indirect representation

Abstract

The inference of phylogenetic trees is one of the most important tasks in computational biology. In this paper, we propose an extension to multi-objective evolutionary algorithms to address this problem. Here, we adopt an enhanced indirect encoding for a tree using the corresponding Prüfer code represented in Newick format. The algorithm generates a range of non-dominated trees given alternative fitness measures such as statistical likelihood and maximum parsimony. A key feature of this approach is the preservation of the evolutionary hierarchy between species. Preliminary experimental results indicate that our model is capable of generating a set of optimized phylogenetic trees for given species data and the results are comparable with other techniques. © 2008 Springer Berlin Heidelberg.