RAxML

RAxML is a fast program for the inference of phylogenies with maximum likelihood method. RAxML offers several evolutionary models for both DNA and amino acid sequences.

RAxML

License

Free to use and open source under GNU GPLv3.

Available

Roihu-CPU: 8.2.12, via the bio-apps module.

Usage

RAxML is part of the bio-apps collection on Roihu. Load the bio-apps module tree and then the RAxML module:

module load bio-apps/v202603
module load raxml/8.2.12

To see the installed RAxML versions, use the command:

module spider raxml

Which version to use?

RAxML is installed in serial and MPI versions, each with generic, SSE3- and AVX-optimized binaries: raxmlHPC, raxmlHPC-SSE3, raxmlHPC-AVX, raxmlHPC-MPI, raxmlHPC-MPI-SSE3 and raxmlHPC-MPI-AVX.

The serial version (raxmlHPC) is intended for small to medium datasets and for initial experiments to determine appropriate search parameters.

The MPI version (raxmlHPC-MPI) is for executing really large production runs (i.e. 100 or 1,000 bootstraps). You can also perform multiple inferences on larger data sets in parallel to find a best-known ML tree for your data set. Finally, the rapid BS algorithm and the associated ML search have also been parallelized with MPI.

The current MPI version only works properly if you specify the number of runs in the command line, since it has been designed to do multiple inferences or rapid/standard BS (bootstrap) searches in parallel. For all remaining options, the usage of this type of coarse-grained parallelism does not make much sense. Please use the -N option instead of the -# option as the latter can be mistaken for a start of a comment by the batch job system.

The AVX-optimized binaries (raxmlHPC-AVX, raxmlHPC-MPI-AVX) can run faster than the non-optimized versions, but can cause problems on some datasets. In case of problems, try the non-optimized or SSE3-optimized versions.

For details, please refer to the chapter "When to use which Version?" in the RAxML manual.

Example batch job scripts

Serial versionMPI version

#!/bin/bash
#SBATCH --account=<project>
#SBATCH --job-name=raxml_serial
#SBATCH --partition=small
#SBATCH --time=10:00:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=8G

module load bio-apps/v202603
module load raxml/8.2.12

raxmlHPC-AVX -s alg -m GTRGAMMA -p 12345 -n test1

#!/bin/bash
#SBATCH --account=<project>
#SBATCH --job-name=raxml_mpi
#SBATCH --partition=small
#SBATCH --time=10:00:00
#SBATCH --ntasks=100
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=2G

module load bio-apps/v202603
module load raxml/8.2.12

srun raxmlHPC-MPI-AVX -N 100 -s cox1.phy -m GTRGAMMAI -p 12345 -n test2

The MPI example above runs 100 ranks within the single-node small partition. For larger, multi-node runs you can use the large partition, which requires a completed scalability test.

Support

CSC Service Desk