BayeScan
BayeScan aims at identifying candidate loci under natural selection from genetic data, using differences in allele frequencies between populations. The analysis is based on the multinomial-Dirichlet model.
License
Free to use and open source under GNU GPLv3
Available
- Puhti: 2.1
Usage
To use BayeScan, first run command
module load biokit
After that you can launch BayeScan with a command like:
bayescan_2.1 -threads 1 test_binary_AFLP.txt
With bayescan_2.1, it is important to define the number of threads always explicitly. This is because, by default, BayeScan tries to use all available cores.
On Puhti, BayeScan tasks should be executed as batch jobs. Below is a sample batch job file for BayeScan:
#!/bin/bash
#SBATCH --job-name=bayescan
#SBATCH --account=project_XXXXXX
#SBATCH --time=08:00:00
#SBATCH --mem=6G
#SBATCH --partition=small
#SBATCH --cpus-per-task=4
#SBATCH --nodes=1
#SBATCH --ntasks=1
module load biokit
bayescan_2.1 -threads ${SLURM_CPUS_PER_TASK} test_binary_AFLP.txt > bayescan_omp.out
The script above reserves 8 hours of computing time, 6 GB of memory and 4 computing cores. The XXXXXX in the --account definition
should be replaced with the ID number of your computing project. The job can be submitted to the batch job system with command:
sbatch script
Don't use BayeScan with more than 8 cores (except if you have verified that your task really benefits from larger core numbers).
More instructions for running batch jobs can be found form CSC batch job instructions