RELION cluster execution#

The following code outlines how to use RELION on a SLURM cluster. Make sure to adapt the parameters to your specific use case.

#!/bin/bash
#SBATCH -p gpu-el8
#SBATCH -J Ref6
#SBATCH --ntasks=15
#SBATCH -o run.out
#SBATCH -e run.err
#SBATCH --open-mode=append
#SBATCH --gres=gpu:2
#SBATCH -C gpu=3090
#SBATCH -N 1
#SBATCH --mem=60GB
#SBATCH --time=1-00:00:00
#SBATCH --export=NONE
#SBATCH --ntasks-per-core=1


module purge
module load RELION/4.0.1-EMBLv.0011_20230510_01_e16f796_a-foss-2022a-CUDA-11.7.0

mpirun \
        -n 3 `which relion_refine_mpi` \
        --i ctf_test.star \
        --ref template52.mrc \
        --o ./run_1/ \
        --ctf \
        --auto_refine \
        --split_random_halves \
        --firstiter_cc \
        --ini_high 60 \
        --dont_combine_weights_via_disc \
        --pool 3 \
        --pad 2 \
        --particle_diameter 250 \
        --flatten_solvent \
        --zero_mask \
        --oversampling 1 \
        --healpix_order 2 \
        --auto_local_healpix_order 4 \
        --offset_range 5 \
        --offset_step 2 \
        --sym C1 \
        --low_resol_join_halves 40 \
        --norm \
        --scale \
        --j 7 \
        --gpu

Assuming you copied the code above to a file run_relion_refine.sbatch you scan submit that to the cluster via:

sbatch run_relion_refine.sbatch
Show Source