Difference between revisions of "AlphaFold2"

Latest revision as of 10:36, 28 September 2023

AlphaFold2 predicts a protein's 3D folding structure by its amino acid sequence with the accuracy that is competitive with experimental results. This AI-powered structure prediction of AlphaFold2 has been recognized as the scientific breakthrough of the year 2021. The AlphaFold package is now installed in the new software stack on Euler.

Changelog

12/09/2023 - Branch for the new script using AlphaFold 2.3.1 merged with main branch and available for all users

3/08/2023 - Uniref90 has been updated

25/07/2023 - New branch of the [ https://gitlab.ethz.ch/sis/alphafold_on_euler alphafold helper script] is currently being tested. This branch uses AlphaFold 2.3.1 and is fully migrated to SLURM.

20/07/2023 - Updated bfd, mgnify, pdb, uniprot and uniref30 databases. Uniref90 is in the process of being updated

17/07/2023 - AlphaFold 2.3.1 is available on Euler. Release notes are available here

Create a job script

A job script is a BASH script containing commands to request computing resources, set up the computing environment, run the application and retrieve the results. Here we propose a breakdown of a typical job script for Alphafold2 on Euler. Please note that you can generate this script by using our custom script available here.

Request computing resources

AlphaFold2 can run with CPUs only, or with CPUs and GPUs which helps speed up the computation significantly. Here we request 8 CPU cores, in total 240GB of memory, 120GB of local scratch space and one GPU. Your SLURM script should start with #!/usr/bin/bash (the shebang) and the #SBATCH pragmas, that detail, line by line, which resources you would like to request for your alphafold run :

#!/usr/bin/bash
#SBATCH -n 8                                                     # Number of CPUs
#SBATCH --time=24:00:00                                          # Runtime
#SBATCH --mem-per-cpu=30000                                      # CPU memory per CPU core
#SBATCH --nodes=1                                                # All CPUs in the same host
#SBATCH -G 1                                                     # Number of GPUs
#SBATCH --gres=gpumem:10240                                      # GPU memory
#SBATCH --tmp=120000                                             # Scratch space per CPU core
#SBATCH -A es_share                                              # Shareholder group name
#SBATCH -J alphafold                                             # Job name

Set up a computing environment for AlphaFold

source /cluster/apps/local/env2lmod.sh
module load gcc/6.3.0 openmpi/4.0.2 alphafold/2.3.1
source /cluster/apps/nss/alphafold/venv_alphafold_2.3.1/bin/activate

Enable Unified Memory (if needed)

If the input protein sequence is too large for a single GPU memory (approximately larger than 1500aa), enable Unified Memory to bridge the system memory to the GPU memory so that you can oversubscribe the GPU memory of a single GPU.

export TF_FORCE_UNIFIED_MEMORY=1
export XLA_PYTHON_CLIENT_MEM_FRACTION="4.0"

Define paths

# Define paths to databases, fasta file and output directory
DATA_DIR="/cluster/project/alphafold" #Path to all of the alphafold databases on the cluster
FASTA_DIR="/cluster/home/jarunanp/fastafiles" #Path to where the fastafile is stored
OUTPUT_DIR=${TMPDIR}/output #Path to the immediate output of the run (in the automatically-generated script it would be the local scratch)

For the output directory, there are two options.

• Use $SCRATCH (max 2.7TB), $HOME (max. 20GB) or group storage (/cluster/project or /cluster/work), e.g.,

OUTPUT_DIR=${SCRATCH}/protein_name/output

• Use the local scratch as the output directory. To do so, request the scratch space with #SBATCH options (e.g., in this example we are requesting 120GB local scratch space in total using the --tmp option). At the end of the computation, don't forget to copy the result from there.

OUTPUT_DIR=${TMPDIR}/output
...
python /path/run_alphafold.py ...
...
cp ${TMPDIR}/output /to/desired/location
or
rsync -av  $TMPDIR/output/ /to/desired/location

Call Python run script

python /cluster/apps/nss/alphafold/alphafold-2.1.1/run_alphafold.py \
--data_dir=$DATA_DIR \
--output_dir=$OUTPUT_DIR \
--max_template_date="2021-12-06" \
--bfd_database_path=$DATA_DIR/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt \
--uniref90_database_path=$DATA_DIR/uniref90/uniref90.fasta \
--uniclust30_database_path=$DATA_DIR/uniclust30/uniclust30_2018_08/uniclust30_2018_08 \
--mgnify_database_path=$DATA_DIR/mgnify/mgy_clusters_2018_12.fa \
--template_mmcif_dir=$DATA_DIR/pdb_mmcif/mmcif_files \
--obsolete_pdbs_path=$DATA_DIR/pdb_mmcif/obsolete.dat \

Then, define the input fasta file, select the model preset (monomer or multimer) and define the path to structure databases accordingly.

• For a monomeric protein

--fasta_paths=$FASTA_DIR/some_protein.fasta \
--model_preset=monomer \
--pdb70_database_path=$DATA_DIR/pdb70/pdb70

• For a multimeric protein

--fasta_paths=$FASTA_DIR/some_complicated_protein.fasta \
--model_preset=multimer \
--pdb_seqres_database_path=$DATA_DIR/pdb_seqres/pdb_seqres.txt \
--uniprot_database_path=$DATA_DIR/uniprot/uniprot.fasta

Enable relaxation on GPU (version >= 2.1.2)
In this version, it is possible to enable running relaxation on GPU with the option --use_gpu_relax. Please see above how to start MPS to use this option.

--use_gpu_relax=1

Submit a job

For SLURM, submit a job with the command

$ sbatch < run_alphafold.sbatch

The screen output will be save in the slurm-JobID.out file, e.g slurm-3435300.out, unless other names for the standard output/error files has been defined with #SBATCH pragmas at the beginning of the script.

From our benchmark, it took around 40 minutes to fold Ubiquitin[76aa] and 2.5 hours to fold T1050[779aa].

Setup script

This setup script creates a job script with estimate computing resources depending on the input protein sequence. To download the setup script:

git clone https://gitlab.ethz.ch/sis/alphafold_on_euler.git

Usage:

./setup_alphafold_run_script.sh -f [Fasta file] -w [work directory] --max_template_date yyyy-mm-dd -b [LSF/SLURM]

Example:

$ ./setup_alphafold_run_script.sh -f ../../fastafiles/IFGSC_6mer.fasta -w $SCRATCH
 Reading /cluster/home/jarunanp/alphafold_run/fastafiles/IFGSC_6mer.fasta
 Protein name:            IFGSC_6mer
 Number of sequences:     6
 Protein type:            multimer
 Number of amino acids:
                   sum: 1246
                   max: 242
 Estimate required resources:
   Run time: 24:00
   Number of CPUs: 12
   Total CPU memory: 120000
   Number of GPUs: 1
   Total GPU memory: 20480
   Total scratch space: 120000
 Output an LSF run script for AlphaFold2: /cluster/scratch/jarunanp/run_alphafold.bsub

For SLURM, submit the script with the command

$ sbatch < run_alphafold.sbatch

Postprocessing

Similar plots as generated by the Colabfold jupyter notebook can be created by the alphafold-postprocessing python script. It is available on Euler as a module

module load gcc/6.3.0 alphafold-postprocessing
postprocessing.py -o plots/ work_directory/

The above command will process pkl files generated by alphafold in the folder work_directory/ and put the resulting plots into a folder plots/.

The postprocessing is integrated in the setup script described above.

Databases

The AlphaFold databases are available for all cluster users at /cluster/project/alphafold.

If you wish to download databases separately, you can see the instruction here.

Example

The Ubiquitin fastafile is provided with the AlphaFold setup script. It can be used to test AlphaFold2 on Euler. If the working directory is on $SCRATCH, a successful run would complete in ~40 min (depending on the type of resources allocated by the batch system) and generate the following files :

Ubiquitin.done
Ubiquitin.out
Ubiquitin.err

Ubiquitin
├── features.pkl
├── msas
│   ├── bfd_uniclust_hits.a3m
│   ├── mgnify_hits.sto
│   ├── pdb_hits.hhr
│   └── uniref90_hits.sto
├── ranked_0.pdb 
├── ranked_1.pdb
├── ranked_2.pdb
├── ranked_3.pdb
├── ranked_4.pdb
├── ranking_debug.json
├── relaxed_model_1_pred_0.pdb
├── relaxed_model_2_pred_0.pdb
├── relaxed_model_3_pred_0.pdb
├── relaxed_model_4_pred_0.pdb
├── relaxed_model_5_pred_0.pdb
├── result_model_1_pred_0.pkl
├── result_model_2_pred_0.pkl
├── result_model_3_pred_0.pkl
├── result_model_4_pred_0.pkl
├── result_model_5_pred_0.pkl
├── timings.json
├── unrelaxed_model_1_pred_0.pdb
├── unrelaxed_model_2_pred_0.pdb
├── unrelaxed_model_3_pred_0.pdb
├── unrelaxed_model_4_pred_0.pdb
└── unrelaxed_model_5_pred_0.pdb

Further readings

• DeepMind Blog post: "AlphaFold: a solution to a 50-year-old grand challenge in biology"
• ETH News: "Computer algorithms are currently revolutionising biology"
• AlphaFold2 presentation slides 21 March 2022
• Downloading AlphaFold databases and benchmark results