Phylogenetic analysis using maximum likelihood methods in homogeneous parallel environments

This paper describes a parallel application for phylogenetic analysis of DNA sequences. The program extends the phylogenetics program Trexml to increase the speed of this rather exhaustive tree searching program. The program uses the Advanced Stepwise Addition Algorithm (ASA) and provides two main advantages over existing applications. The size of the tree space can be chosen in such a way that we can fully utilize super computing resources. We also adopt a Single Program multiple Data (SPMD) approach in our program and use deterministic means to evaluate which processors are involved in communication. This approach alleviates the communication costs incurred in a master/slave system. Results from running the program on the AC3 Barossa supercomputer indicate that our approach scales well.