Bayesian Evolutionary Analysis Sampling Trees is a cross-platform program for Bayesian analysis of molecular sequences using MCMC. The program is orientated toward (strict and relaxed) molecular clock analyses. It can be used as a method of constructing phylogenies, but it is also intended for testing evolutionary hypotheses without conditioning on a single tree topology. BEAST uses MCMC to average over tree space, so that each tree is weighted proportional to its posterior probability. It uses an XML input format that allows the user to design and run a large range of models. We also include a program that can convert NEXUS files into this format.
The BEAST home page is at http://beast2.org.
Packages with modules
BEAST is released at no cost under the terms of the GNU Lesser General Public Licence.
Example script for the Pan cluster
#!/bin/bash -e #SBATCH --job-name MyBEASTJob #SBATCH --account nesi99999 #SBATCH --time 01:00:00 #SBATCH --cpus-per-task 16 #SBATCH --mem-per-cpu 4G #SBATCH --output MyBEASTJob.%j.out # Include the job ID in the names #SBATCH --error MyBEASTJob.%j.err # of the output and error files module load BEAST/2.2.1 # Here Java is told it can use ( cpus-per-task * mem-per-cpu - 2GB ) = 62GB export _JAVA_OPTS="-Xms62g -Xmx62g" srun beast -beagle -beagle_SSE -threads $SLURM_CPUS_PER_TASK input.xml
Please note the Java options in the above script. BEAST is a Java program, and Java will not automatically use all the memory which Slurm has allocated for it unless told to do so.
Long BEAST runs
If your Slurm job ends too early you can continue its calculation in a second job by way of the
-resume option of BEAST2. For very long BEAST runs it may even make sense to deliberately break it into 1-week segments to get around our limits on maximum job duration.