High-Performance Phylogeny Reconstruction Under Maximum Parsimony

The similarity of the molecular matter of the organisms on Earth suggest that they all share a common ancestor. Thus any set of species is related, and this relationship is called a phylogeny. The links (or evolutionary relationships) among a set of organisms (or taxa) form a phylogenetic tree, where modern organisms are placed at the leaves and ancestral organisms occupy internal nodes, with the edges of the tree denoting evolutionary relationships. Scientists are interested in evolutionary trees for the usual reasons of scientific curiosity. However, phylogenetic analysis is not just an academic exercise. Phylogenies are the organizing principle for most biological knowledge. As such, they are a crucial tool in identifying emerging diseases, predicting disease outbreaks, and protecting ecosystems from invasive species [Bader et al., 2001, Cracraft, 2002]. The greatest impact of phylogenetics will be reconstructing the Tree of Life, the evolutionary history of all-known organisms. No one precisely knows the number of organisms that exist in the world. Estimates often cited range from 10-100 million species. Today, only about 1.7 million species are known, and a very small fraction (i.e., on the order of 0.4%) are included in any sort of phylogenetic tree [Yates et al., 2004]. Given the societial benefits of phylogenetic trees, reconstructing the evolutionary history from present-day taxa is a very difficult problem. For n organisms, there are (2n − 5)(2n − 3) · · · (5)(3) distinct binary trees; each a possible hypothesis for the " true " evolutionary history. (There are over 13 billion possible trees for 13 taxa.) Since the size of the tree space increases exponentially with the number of taxa, it impossible to explore all possible hypothesis for them within a reasonable time frame. Most phylogenetic methods limit themselves to exhausitive searches on extremely small datasets (< 50 sequences) or heuristic strategies for larger datasets. Another difficulty lies in accessing the accuracy of the reconstructed tree. Short of traveling back into time, there is no way of determining whether the proposed evolutionary history is 100% correct. The best conclusion is simply the best hypothesis of what might have happened. There is an ongoing search for tree-building methods that are the most robust (i.e., that are more likely to estimate the true topology even when the evolutionary assumptions are violated), consistent (that converge on the true topology as more data are added), and efficient (that converge on the topology most quickly). …