PhyML (version 3.3.20190909)

Alignment file:

phylip format

Changes interleaved format:

Number of data sets:

Data type:

Transition/transversion ratio:

Must be a positive integer, 'e' if you want PhyML to estimate it

Substitution model:

Proportion of invariant sites:

Can be a fixed value in the [0,1] range or 'e' to get the maximum likelihood estimate, 0 to ignore this parameter

Equilibrium frequencies:

Empirical: frequencies are estimated by counting the occurences in the alignment. ML/Model: frequencies are estimated using ML for nucleotides or defined by the proteic substitution model.

Number of categories for the discrete gamma model:

1 means no gamma model

Parameter of the gamma model:

'e' if you want PhyML to estimate it

Tree topology search:

Optimise topology:

Statistical tests for branch support:

Use aLRT or aBayes to save computing time

Number of seed used to initiate the random number generator:

0 to random seed

Tree file:

newick format

PhyML 3.3

A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood''

Overview:

PhyML is a phylogeny software based on the maximum-likelihood principle. Early PhyML versions used a fast algorithm to perform Nearest Neighbor Interchanges (NNIs), in order to improve a reasonable starting tree topology. Since the original publication (Guindon and Gascuel 2003), PhyML has been widely used due to its simplicity and a fair accuracy/speed compromise. In the mean time research around PhyML has continued.

We designed an efficient algorithm to search the tree space using Subtree Pruning and Regrafting (SPR) topological moves (Hordijk and Gascuel 2005), and proposed a fast branch test based on an approximate likelihood ratio test (Anisimova and Gascuel 2006). However, these novelties were not included in the official version of PhyML, and we found that improvements were still needed in order to make them effective in some practical cases. PhyML 3.0 achieves this task.

It implements new algorithms to search the space of tree topologies with user-defined intensity. A non-parametric, Shimodaira-Hasegawa-like branch test is also available. The program provides a number of new evolutionary models and its interface was entirely re-designed. We tested PhyML 3.0 on a large collection of real data sets to ensure that the new version is stable, ready-to-use and still reasonably fast and accurate.

For further informations, please visite the PhyML website.

Please cite this paper if you use this software in your publications.