Click here to see an image of the PHYML Web site. Top left: Stephane Guindon, top right: Olivier Gascuel. Bottom left: animated image represents the "growth of phylogenetic tree."

Phylogenetics has a wide number of applications, from functional genomics to the study of biodiversity. Phylogenetics becomes highly topical with the emerging phylogenomic and the "Tree of Life" projects, as it provides invaluable information on the relationships between species, genomes, organelles, and genes. The number of genetic sequences in databases that can be used for such studies is now considerable. Data sets with more than a hundred sequences and ten thousand sites are quite common, and estimating phylogenetic trees from these very large amounts of data is challenging from a computational perspective. Our paper presents a simple, but fast and accurate algorithm that addresses this problem. We also provide user friendly software, PHYML, which implements this algorithm. People are starting to use (and cite) PHYML as this tool is well suited for large-scale phylogenetic studies.

PHYML is based on the maximum-likelihood principle, which is commonly acknowledged to be the most accurate approach to infer phylogenies. Traditional maximum-likelihood methods were slow. Our new algorithm is one order of magnitude faster than the standard approach. Trees involving hundreds of sequences with thousands of sites can now be inferred in a few minutes using a simple personal computer. This was impossible with previous, standard maximum-likelihood computer programs.

Our paper describes a method to estimate the evolutionary relationships between organisms from the comparison of their DNA or amino acid sequences. The strength of this approach lies in its speed which makes possible the analysis of large amount of data in reasonable computing time. This method is also based on statistical models that help us to better describe and understand the mechanisms of molecular evolution.