Accurate structure prediction of biomolecular interactions with AlphaFold 3
Problem
Predicting the joint 3D structures of biomolecular complexes (proteins with ligands, nucleic acids, ions and antibodies) previously required separate specialized tools with limited accuracy. AlphaFold 3 addresses this unified modelling problem.
Why it matters
Accurate models of biomolecular interactions underpin drug discovery and the understanding of disease mechanisms at the molecular level.
What the AI did
AlphaFold 3, a diffusion-based deep learning model, jointly predicts structures of complexes containing proteins, nucleic acids, small molecules, ions and modified residues, outperforming prior specialized tools.
Results
Substantially higher accuracy than state-of-the-art docking tools for protein-ligand, nucleic-acid-specific predictors for protein-nucleic acid, and AlphaFold-Multimer v2.3 for antibody-antigen; no numeric metrics in abstract.
Impact
Peer-reviewed in Nature with very high citations, showing high-accuracy modelling across biomolecular space is possible in one model.
Record
| Venue | Nature |
|---|---|
| Published | 2024 |
| Work type | article |
| Authors | Josh Abramson, Jonas Adler, Jack Dunger, Richard Evans, Tim Green, Alexander Pritzel, Olaf Ronneberger, Lindsay Willmore et al. |
| Citations | 15,361 (OpenAlex, at import time) |
| References | 78 |
| Model or method | AlphaFold 3 (diffusion-based architecture) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41586-024-07487-w |
| Open access | hybrid |
| Keywords | AlphaFold 3, biomolecular interactions, protein-ligand, antibody-antigen, diffusion model, deep learning |
Importance rating
major — millions of lives, an industry or a field
Abstract states AlphaFold 2 spurred a revolution in protein structure modelling enabling a huge range of applications, and AlphaFold 3 unifies high-accuracy modelling across biomolecular space; major impact on a foundational field.