NEWS
08 May 2024
Latest version of the AI models how proteins interact with other molecules — but DeepMind restricts access to the tool.
By Ewen Callaway
Since the powerhouse artificial intelligence (AI) tool AlphaFold2 was released in 2021, scientists have used the protein-structure-prediction model to map one of our cells’ biggest machines, discover drugs and chart the universe of every known protein.
Despite such successes, John Jumper — who leads AlphaFold’s development at Google DeepMind in London — is regularly asked whether the tool can do more. Requests include the ability to predict the shape of proteins that contain function-altering modifications, or their structure alongside those of DNA, RNA and other cellular players that are crucial to a protein’s duties. “I would say ‘no, you can’t put that into AlphaFold’,” Jumper says. “I would rather solve their problems.”
The latest version of AlphaFold, described on 8 May in Nature [1], aims to do just that — by giving scientists the ability to predict the structures of proteins during interactions with other molecules. But whereas DeepMind made the 2021 version of the tool freely available to researchers without restriction, AlphaFold3 is limited to non-commercial use through a DeepMind website.
Frank Uhlmann, a biochemist at the Francis Crick Institute in London who gained early access to AlphaFold3, has been impressed with its capabilities. “This is just revolutionary,” he says. “It’s going to democratize structural-biology research.”
[1] Abramson, J., Adler, J., Dunger, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature (2024). https://doi.org/10.1038/s41586-024-07487-w
Comments