AlphaFold3
1. Introduction
This package provides an implementation of the inference pipeline of AlphaFold 3. See below for how to access the model parameters. You may only use AlphaFold 3 model parameters if received directly from Google. Use is subject to these terms of use.
Any publication that discloses findings arising from using this source code, the model parameters or outputs produced by those should cite the Accurate structure prediction of biomolecular interactions with AlphaFold 3 paper.
Please also refer to the Supplementary Information for a detailed description of the method.
AlphaFold 3 is also available at alphafoldserver.com for non-commercial use, though with a more limited set of ligands and covalent modifications.
If you have any questions, please contact the AlphaFold team at alphafold@google.com.
1.1 Local Installation
The CNCA team prepared a local installation of AlphaPhold3 using a container based on singularity or (appatainer) including the Genetic Database.
The Model Parameters is not included, users must request access filling a form and comply with the Google DeepMind terms of use at all times.
The local installation provide the AlphaFold3 version 3.0.1 over a container based on Ubuntu 24.04 distribution with cuda-12.6. The container is already prepared and is used throughtout a wrapper that accept all run_alphafold.py options; the Genetic Database location is already configured, users should provide the json path, model and output directories location.
The main resource target of AlphaFold3 is the GPU but the application can execute the data stage only on the CPU although the performance is substantially worst. The model inference stage requires a GPU, if you submit a job to a partition without a GPU, such as the hpc or fct partitions, then the option --norun-inference is added automatically the the run_alphafold.py script.
The user is free to prepare is own container as explained on https://github.com/google-deepmind/alphafold3/tree/main.
1.2 Environment
The environment is activate with command
$ module load alphaphold3/3.0.1
this will activate automatically a virtual environment ready to start the AlphaFold3 container throught the python script run_alphafold.py.
1.3 Data Base Location
The Genetic Database is installed only on local disk on workernodes bellow the directory
/local/alphafold3
The environment variable ALPHAFOLD3_DATABASE points to this location.
The Genectic Database is not available on the user interfaces.
2. How to Run
2.1 Choose a working directory
$ mkdir ~/alphafold3_test $ cd ~/alphafold3_test
2.1 Obtain the Model Parameters
The model parameters access is subject to Google DeepMind terms of use, please fill the form https://forms.gle/svvpY4u2jsHEwWYS6 and follow the instructions. Once you have access to the model parameters you will be responsible to keep the data private.
2.1 Choose a working directory
2.1 Choose a working directory
2.1 Choose a working directory
2.1 Choose a working directory
2.1 Choose a working directory
2.1 Choose a working directory
Upgrades may be requested using the helpdesk@incd.pt address.