Alpha Fold 3

This package provides an implementation of the inference pipeline of AlphaFold 3. We have installed the tool as a module alphafold/3.0.0.

module load alphafold/3.0.0

Obtaining Genetic Databases

This step requires wget and zstd to download the Genetic Databases. We have downloaded the databases under /ix1/genomics/refs/alphafold/alphafold3.

Obtaining Model Parameters

To request access to the AlphaFold 3 model parameters, please complete this form. Access will be granted at Google DeepMind’s sole discretion. Google DeepMind will aim to respond to requests within 2–3 business days. You may only use AlphaFold 3 model parameters if received directly from Google. Use is subject to these terms of use.

Download af3.bin.zst to a folder models.

AlphaFold 3 Input

You can provide inputs to run_alphafold.py in one of two ways:

• Single input file: Use the --json_path flag followed by the path to a single JSON file.
• Multiple input files: Use the --input_dir flag followed by the path to a directory of JSON files.

AlphaFold 3 uses a custom JSON input format.

This is an example input: /ix1/bioinformatics/alphafold3/input.json

Running Your Prediction

This is an example batch job: /ix1/bioinformatics/alphafold3/job.sbatch

#!/bin/bash
#SBATCH --job-name=alphafold3
#SBATCH -M gpu
#SBATCH --partition=l40s
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=16
#SBATCH --mem=256GB
#SBATCH --gres=gpu:1
#SBATCH --time=1-00:00:00

module load alphafold/3.0.0

mkdir -p output

run_alphafold.py \
     --json_path=input.json \
     --model_dir=models \
     --db_dir=/ix1/genomics/refs/alphafold/alphafold3 \
     --output_dir=output

To submit the job, sbatch job.sbatch.