Parallel divide-and-conquer phylogeny reconstruction by maximum likelihood

Abstract

Phylogenetic trees are important in biology since their applications range from determining protein function to understanding the evolution of species. Maximum Likelihood (ML) is a popular optimization criterion in phylogenetics. However, inference of phylogenies with ML is NP-hard. Recursive-Iterative-DCM3 (Rec-I-DCMS).is a divideand-conquer framework that divides a dataset into smaller subsets (subproblems), applies an external base method to infer subtrees, merges the subtrees into a comprehensive tree, and then refines the global tree with an external global method. In this study we present a novel parallel implementation of Rec-I-DCM3 for inference of large trees with ML. Parallel-Rec-I-DCM3 uses RAxML as external base and global search method. We evaluate program performance on 6 large real-data alignments containing 500 up to 7.769 sequences. Our experiments show that P-Rec-I-DCM3 reduces inference times and improves final tree quality over sequential Rec-I-DCM3 and stand-alone RAxML. © Springer-Verlag Berlin Heidelberg 2005.