Genetics · Prediction and risk
AlphaGenome: AI that predicts how DNA changes might affect gene activity (research tool)
In January 2026 Google DeepMind published AlphaGenome in Nature: a model that reads up to a million letters of DNA and predicts how small changes (variants) affect how genes are switched on and off. It's a research tool for scientists and isn't used to diagnose patients.
How far away is this?
It's a peer-reviewed research model evaluated on benchmarks. It isn't a medical test and hasn't been shown to improve patient diagnosis. Predicting a change in gene activity is not the same as predicting disease.
Where is it approved?
| Region | Regulator | Status | Notes |
|---|---|---|---|
| United States | FDA | Research only | A research tool, not a medical product. |
| European Union | EUDAMED | Research only | A research tool, not a medical product. |
| United Kingdom | MHRA | Research only | A research tool, not a medical product. |
| Australia | TGA | Research only | A research tool, not a medical product. |
A research tool for scientists. It isn't a medical product, so there's no regulatory approval to track. "Register not searched yet" means we haven't searched that regulator's register; it doesn't mean the product is or isn't available. Approval in one country isn't approval in another.
What is it?
Most of our DNA doesn't make proteins but controls when genes turn on. Spotting which tiny spelling changes in those regions matter is hard. AlphaGenome predicts thousands of molecular measurements from DNA sequence alone, and how a variant would change them.
What's the evidence?
AlphaGenome takes 1 Mb of DNA sequence as input and predicts thousands of functional genomic tracks, up to single-base-pair resolution. Published in Nature, 28 January 2026.[1]
Evidence stage: Early research · #c1
It matched or beat the strongest available external models in 25 of 26 evaluations of variant effect prediction. These are the developers' own benchmark comparisons.[1]
Evidence stage: Early research · #c2
Independent commentary in Nature Structural & Molecular Biology called it a step forward but said there's still room for improvement.[2]
Evidence stage: Early research · #c3
What does it mean for me?
Nothing changes for you today. In the longer term, tools like this could help genetics labs make sense of unexplained results, such as a 'variant of uncertain significance'. Don't put your DNA data into any online tool claiming to diagnose you with it.
What could I ask my doctor?
- My genetic test found a 'variant of uncertain significance'. Will it be re-checked as the science improves?
- Who reviews my genetic results, and how?
These are general starting points, not a recommendation to choose any particular test or treatment.
Change history
- : First published.
- : Updated after AI-assisted source check (originally mislabelled as independent fact-check and clinical review; corrected below).
- : 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
- Nature: Advancing regulatory variant effect prediction with AlphaGenome. nature.com. Peer-reviewed study. https://www.nature.com/articles/s41586-025-10014-0
- Nature Structural & Molecular Biology: News & Views. nature.com. Expert commentary. https://www.nature.com/articles/s41594-026-01763-1
Publisher shown as the website hosting each source. Dates are given where the source label includes them.