alphafold_singularity

Singularity recipe for AlphaFold, with example Slurm job script.

This splits off my pull request google-deepmind/alphafold#166

Disclaimer: this project is not affiliated with DeepMind.

IMPORTANT - main branch not usable directly

The main branch here is not directly usable with the main branch of AlphaFold. This Singularity recipe works only with the matching release of AlphaFold. Please use one of the releases here, with a matching release of AlphaFold (ignoring the bugfix number).

Unfixed bug in Alphafold tagged release 2.2.2

N.B. google-deepmind/alphafold#510 (comment)

Prebuilt Singularity image

A prebuilt image is hosted on cloud.sylabs.io: https://cloud.sylabs.io/library/prehensilecode/alphafold_singularity/alphafold

What This Code Contains

• Singularity.def which is the recipe to build the Singularity image. This is a port of the Dockerfile provided by AlphaFold.
• run_singularity.py which is a port of the run_docker.py script provided by AlphaFold. It is a wrapper to provide a friendly interface for running the container.

Installation Instructions

Download AlphaFold and alphafold_singularity:

N.B. The AlphaFold version and the alphafold_singularity versions must match, except for the bugfix number in alphafold_singularity. E.g. alphafold_singularity 2.3.2-1 goes with AlphaFold 2.3.2

$ export ALPHAFOLD_VERSION=2.3.2
$ wget https://github.com/deepmind/alphafold/archive/refs/tags/v${ALPHAFOLD_VERSION}.tar.gz -O alphafold-${ALPHAFOLD_VERSION}.tar.gz
...
2023-02-08 17:28:50 (1.24 MB/s) - ‘alphafold-x.x.x.tar.gz’ saved [5855095]
$ tar -xvf alphafold-${ALPHAFOLD_VERSION}.tar.gz
$ cd alphafold-${ALPHAFOLD_VERSION}
$ wget https://github.com/prehensilecode/alphafold_singularity/archive/refs/tags/v${ALPHAFOLD_VERSION}.tar.gz -O alphafold_singularity-${ALPHAFOLD_VERSION}.tar.gz
...
2023-02-08 17:42:18 (1.58 MB/s) - ‘alphafold_singularity-x.x.x.tar.gz’ saved [10148]
$ tar -xf alphafold_singularity-${ALPHAFOLD_VERSION}.tar.gz
$ mv alphafold_singularity-${ALPHAFOLD_VERSION} singularity
$ python3 -m pip install -r singularity/requirements.txt
$ sudo singularity build alphafold.sif singularity/Singularity.def

Build the Singularity image

First install the Python requirements:

$ python3 -m pip install -r singularity/requirements.txt

Then, build the Singularity image:

$ sudo singularity build alphafold.sif singularity/Singularity.def

If your /tmp directory is small, you may need to set the SINGULARITY_TMPDIR environment variable to a directory on a filesystem with more free space. My builds have consumed up to 15 GiB of space. The resulting image file may be up to 10 GiB.

Download genetic databases

See AlphaFold 2.3.2 README for instructions on downloading genetic databases. These are necessary to run AlphaFold.

This step requires aria2c.

N.B. The difference between downloading the "reduced databases" as opposed to the "full databases" is that the reduced databases download "small BFD" instead of "BFD".

Modify run script, install, and run

To run, modify the $ALPHAFOLD_SRC/singularity/run_singularity.py and change the section marked USER CONFIGURATION. At the least, you will need to modify the values of:

• singularity_image - absolute path to the alphafold.sif Singularity image

E.g.

#### USER CONFIGURATION ####
# AlphaFold Singularity image.
singularity_image = Client.load(os.path.join(os.environ['ALPHAFOLD_DIR'], 'alphafold.sif'))

Running on an HPC cluster

Currently, this project only supports Slurm. Please open an issue to request support for other job schedulers/resource managers.

Run as a Slurm job on a cluster

See the example job script example_slurm_job.sh. N.B. this example must be modified to suit your specific HPC environment.

The run_singularity.py script will use all GPUs available to the job. If Slurm has been set up with cgroups, the job may request fewer than the total number of GPUs installed on a node. E.g. if the GPU nodes in the cluster have 4 GPU devices each, the job can do

#SBATCH --gpus=2

and AlphaFold Singularity will use only two of the four GPUs. This is because the cgroup for the job only shows 2 GPUs to the job.