Accurate structure prediction of biomolecular interactions with AlphaFold3
Received: 19 December 2023
Accepted: 29 April 2024
Published online: 8 May 2024
Open access
josh_abramson|Josh Abramson]] ^ 17 Jonas Adler ^ 17 Jack Dunger ^ 1,7 Richard Evans ^ 1,7 Tim Green ^ 1,7 Alexander Pritzel ^ 17 Olaf Ronneberger ^ 17 Lindsay Willmore ^ 1,7 Andrew J Ballard ^ 1 Joshua Bambrick ^ 2 Sebastian W Bodenstein ^ David A Evans ^ Chia-Chun Hung ^ 1 Michael O'Neill ^ David Reiman ^ 1 Kathryn Tunyasuvunakool ^ Zachary Wu ^ Akvilè Žemgulytè ^ Eirini Arvaniti ^ 3 Charles Beattie ^ 3 Ottavia Bertolli ^ 3 Alex Bridgland ^ 3 Alexey Cherepanov ^ 4 Miles Congreve ^ 4 Alexander I Cowen-Rivers ^ 3 Andrew Cowie ^ 3 Michael Figurnov ^ 3 Fabian B Fuchs ^ 3 Hannah Gladman ^ 3 Rishub Jain ^ 3 Yousuf A Khan ^ 3,5 Caroline M R Low ^ 4 Kuba Perlin ^ 3 Anna Potapenko ^ 3 Pascal Savy ^ 4 Sukhdeep Singh ^ 3 Adrian Stecula ^ 4 Ashok Thillaisundaram ^ 3 Catherine Tong ^ 4 Sergei Yakneen ^ 4 Ellen D Zhong ^ 3,6 Michal Zielinski ^ 3 Augustin Židek ^ 3 Victor Bapst ^ 1,8 Pushmeet Kohli ^ 1,8 Max Jaderberg ^ 2,8 ϖ Demis Hassabis ^ 1,2,8 ⊠ John M Jumper ^ 1,8 ⊠
${ }^{1}$ Core Contributor, Google DeepMind, London, UK.
${ }^{2}$ Core Contributor, Isomorphic Labs, London, UK.
${ }^{3}$ Google DeepMind, London, UK.
${ }^{4}$ Isomorphic Labs, London, UK.
${ }^{5}$ Department of Molecular
The introduction of AlphaFold $2^{1}$ has spurred a revolution in modelling the structure of proteins and their interactions, enabling a huge range of applications in protein modelling and design ${ }^{2-6}$. Here we describe our AlphaFold 3 model with a substantially updated diffusion-based architecture that is capable of predicting the joint structure of complexes including proteins, nucleic acids, small molecules, ions and modified residues. The new AlphaFold model demonstrates substantially improved accuracy over many previous specialized tools: far greater accuracy for protein-ligand interactions compared with state-of-the-art docking tools, much higher accuracy for protein-nucleic acid interactions compared with nucleic-acid-specific predictors and substantially higher antibody-antigen prediction accuracy compared with AlphaFold-Multimer v.2.3 ${ }^{7,8}$. Together, these results show that high-accuracy modelling across biomolecular space is possible within a single unified deep-learning framework.