Image: © Сергей Шиманович/Stock.adobe.com
DeepMind claims its AI model Alphafold 3 can boost our understanding of the biological world and improve drug discovery.
Google-owned DeepMind has teamed up with Isomorphic Labs to create the latest version of its AlphaFold AI model – and claims this one could transform our understanding of biology.
The companies claim AlphaFold 3 is able to predict the structure and interactions of all life’s molecules with “unprecedented accuracy”. It is hoped that the discoveries from this model will boost our understanding of the biological world and improve drug discovery.
This is the latest iteration of DeepMind’s landmark AI model, which was used to predict the structure of nearly every protein known to science in 2022 – more than 200m proteins in total.
DeepMind followed this up in 2023 with an updated AlphaFold that it claimed could predict various biological molecules with “atomic accuracy”.
The company claims AlphaFold 3 is a 50pc improvement when it comes to predicting the interactions of proteins with other molecule types, compared to existing prediction methods. For some “important categories of interaction”, DeepMind claims this new model has double the prediction accuracy.
“AlphaFold 3 takes us beyond proteins to a broad spectrum of biomolecules,” DeepMind said in a blogpost. “This leap could unlock more transformative science, from developing biorenewable materials and more resilient crops, to accelerating drug design and genomics research.”
DeepMind said AlphaFold 3 is able to model large biomolecules such as proteins and DNA, but it can also model small molecules known as ligands – the company said this is a category encompassing many drugs.
As part of the collaboration, Isomorphic Labs is using this AI model to improve its drug design procedures, by learning how to approach “new disease targets” and develope novel ways to pursue existing ones that were “previously out of reach”.
“AlphaFold 3 assembles its predictions using a diffusion network, akin to those found in AI image generators,” DeepMind said. “The diffusion process starts with a cloud of atoms, and over many steps converges on its final, most accurate molecular structure.
“AlphaFold 3’s predictions of molecular interactions surpass the accuracy of all existing systems. As a single model that computes entire molecular complexes in a holistic way, it’s uniquely able to unify scientific insights.”
DeepMind has created various AI models and claims they can tackle issues such as materials discovery, weather prediction and more. But one of their claims was put under scrutiny recently. A study analysing one of DeepMind’s research papers claimed there is little evidence the company managed to find novel materials using AI.
Find out how emerging tech trends are transforming tomorrow with our new podcast, Future Human: The Series. Listen now on Spotify, on Apple or wherever you get your podcasts.
