aceGoogle DeepMind launches AI tool to help identify genetic drivers of disease
Researchers at Google DeepMind have unveiled their latest artificial intelligence tool and claimed it will help scientists identify the genetic drivers of disease and ultimately pave the way for new treatments. AlphaGenome predicts how mutations interfere with the way genes are controlled, changing when they are switched on, in which cells of the body, and whether their biological volume controls are set to high or low. Most common diseases that run in families, including heart disease and autoimmune disorders,…
On 28 January 2026, Google DeepMind announced AlphaGenome, an AI model that predicts how mutations in non-coding DNA affect gene regulation. Trained on public human and mouse genetics databases, it can assess up to 1m letters of DNA at once to forecast impacts on gene switching across tissues.
The tool matters because most common familial diseases and many cancers are linked to regulatory mutations that are hard to pinpoint in 98% of the genome. If validated, it could accelerate basic understanding of the genome and help prioritize targets for drugs and gene therapies, though researchers note predictions still need experimental confirmation.
- AlphaGenome was trained on public databases of human and mouse genetics to learn links between mutations and gene regulation in specific tissues.
- The tool can analyze up to 1m letters of DNA at once and predict effects on when, where, and how strongly genes are switched on.
- Researchers suggest it could underpin new gene therapies by designing DNA sequences to switch on a gene in nerve cells but not in muscle cells.
- Details of the work were published in Nature and early users include researchers studying cancer drivers.
AlphaGenome analyzes up to 1m letters of DNA to predict how non-coding mutations alter gene regulation across cell types, helping researchers pinpoint genetic drivers of disease and design targeted DNA sequences.
The rundown
Google DeepMind researchers unveiled AlphaGenome on 28 January 2026, describing a model that predicts how mutations interfere with gene control, including timing, cell-type specificity, and expression level. The human genome is 3bn base pairs, with about 2% coding for proteins and the rest orchestrating regulation.
The team trained the system on human and mouse genetics databases. External researchers cited in the piece include Carl de Boer at University of British Columbia, Marc Mansour at UCL working on paediatric haemato-oncology, and Gareth Hawkes at University of Exeter, who noted the non-coding genome is 98% of the genome.
Sources
- JournalismThe Guardian2026-01-28
How should this claim be treated?
The debate