Bayesian phylogeny on Grid

Grid computing defines the combination of computers orclusters of computers across networks, like the internet, to form a distributedsupercomputer. This infrastructure allows scientists to processcomplex and time consuming computations in parallel on demand. Phylogeneticinference for large datasets of DNA/protein sequences is knownto be computationally intensive. Bayesian algorithms allows the estimationof important parameters on species divergence modus and time butat the price of running repetitive long series of Monte Carlo simulations.As part of the BioinfoGrid project, we ported parallel MrBayes tothe EGEE (Enabling Grids for E-sciencE ) grid infrastructure. As casestudy we utilise this supercomputer and program to investigate both achallenging dataset of arthropod phylogeny and the most appropriatemodel of amino acid replacement for that dataset. We relate in this documentthe parallel usage of Bayesian inference and Metropolis-CoupledMarkov Chain Monte Carlo (MCMCMC) to analyse our dataset of completemitochondrial genomes from the Pancrustacea, aiming at resolvingthe position of basal hexapod lineages with respect to Insecta and Crustacea.In this effort, a new matrix of protein change was derived from thedataset itself, and its performance compared with other currently usedmodels.