Drug discovery · Drug discovery
AlphaFold 3: predicting how the molecules of life fit together
AlphaFold 3, published in Nature in May 2024 by Google DeepMind and Isomorphic Labs, predicts how proteins interact with DNA, RNA and drug-like molecules. Demis Hassabis and John Jumper of Google DeepMind shared half the 2024 Nobel Prize in Chemistry for AlphaFold2. It's widely used in research, but we're not aware of any medicine from it reaching patients yet.
How far away is this?
It's a computer prediction tool. Any medicine designed with its help would still need to go through lab, animal and human trials.
Where is it approved?
| Region | Regulator | Status | Notes |
|---|---|---|---|
| United States | FDA | Not applicable | A research tool, not a medical product. |
| European Union | CE mark / EMA | Not applicable | A research tool, not a medical product. |
| United Kingdom | MHRA | Not applicable | A research tool, not a medical product. |
| Australia | TGA | Not applicable | A research tool, not a medical product. |
A research tool used by scientists worldwide. There are no treatments from it for patients yet. "Unknown" means we haven't verified it, not that it's unavailable. Approval in one country isn't approval in another.
What is it?
Proteins are the machinery of our bodies, and their 3D shape decides what they do. AlphaFold predicts those shapes from their building blocks. Version 3 also predicts how proteins bind to other molecules, which matters for designing drugs.
What's the evidence?
AlphaFold 3 predicts joint structures of proteins, nucleic acids, small molecules, ions and modified residues, and outperformed specialised tools on several interaction types in the authors' tests. The authors work for the companies that built it (Google DeepMind and Isomorphic Labs).[1]
Evidence stage: Early research · #c1
Demis Hassabis and John Jumper of Google DeepMind shared half the 2024 Nobel Prize in Chemistry 'for protein structure prediction' (for AlphaFold2).[2]
Evidence stage: Early research · #c2
Independent benchmarking (FoldBench) found AlphaFold 3 the most accurate on most tasks, but antibody–antigen predictions still fail more than half the time and accuracy drops for molecules unlike its training data.[3]
Evidence stage: Early research · #c3
What does it mean for me?
Nothing changes for your care today. In the longer run, tools like this may shorten the early, lab stage of drug discovery. That's a big 'may', and the slow parts (trials) are still slow.
What could I ask my doctor?
- Are any of my treatment options based on new drug-design methods?
- How long does it usually take for a new drug to reach patients?
These are general starting points, not a recommendation to choose any particular test or treatment.
Change history
- : First published.
- : Updated after independent fact-check and clinical review
- : Review label corrected: this page was not clinically reviewed. Earlier labels saying 'independent fact-check' and 'clinical review' were wrong; checks so far were AI-assisted source checks only.
General information only, not medical advice. Talk to your doctor about your situation. AI-assisted source check. Not clinically reviewed.
References
- Abramson et al., Nature (2024). nature.com (2024). Peer-reviewed study. https://www.nature.com/articles/s41586-024-07487-w
- Nobel Prize in Chemistry 2024 press release. nobelprize.org (2024). Official. https://www.nobelprize.org/prizes/chemistry/2024/press-release/
- FoldBench benchmarking, Nature Communications. nature.com. Peer-reviewed study. https://www.nature.com/articles/s41467-025-67127-3
Publisher shown as the website hosting each source. Dates are given where the source label includes them.