Implementation of a parallel processing system for inference of phylogenetic trees

A parallel processing system for inference of phylogenetic trees is reported. The trees are inferred from DNA sequence data by using a maximum likelihood method based on a stochastic model of the evolutional process. This method gives a good representation of the trees, but the computational cost has hindered the use of this method for inferring trees with more than about 20 species. By a parallel processing method based on a function partitioning approach, the authors reduced the cost significantly and obtained a phylogenetic tree of several hundred species on a massively parallel machine.<
>