RAxML is a highly optimized derivative of fastDNAml that allows for the maximum likelihood-based inference of phylogenetic trees with enormous datasets. For example, the serial version can infer the ML tree from a 25,000 taxa by 1,500 base pair dataset in under two weeks.

Using RAxML on HPC

To run the serial version of RAxML, minimally specify an input file, in PHYLIP format, output file name extension, and your desired model of evolution.  For instance with a protein dataset:

raxml -s subset.phy -n subset.raxml -m PROTMIXJTT

"raxml -h" will produce an option summary. The program name listed in the documentation, raxmlHPC, will also work. See the complete documentation file for all available options and useful suggestions. If you need to perform a bootstrapped analysis see mpiRAxML.