Random tree optimization for the construction of the most parsimonious phylogenetic trees

With the availability of ever-increasing gene sequence data across a large number of species, reconstruction of phylogenetic trees to reveal the evolution relationship among those species becomes more and more important. In this paper, we focus on the construction of the most parsimonious phylogenetic trees given sequence data of a group of species as parsimony is probably the most widely used among all tree building algorithms [4]. The major contribution of this paper is the presentation of a novel algorithm, the random tree optimization (RTO) algorithm based on cross-entropy method [16], for the construction of the most parsimonious phylogenetic trees. We analyze the RTO algorithm in the framework of expectation maximization (EM) and point out the similarities and differences between traditional EM algorithm and the RTO algorithm.