Artificial intelligence is moving from analysing images to decoding the very building blocks of life. Google's DeepMind has unveiled a powerful new AI, AlphaGenome 2, aimed at understanding how tiny changes in our DNA can lead to disease.
A New Tool for a Genetic Mystery
Scientists have
long known that our DNA holds the blueprint for our bodies, but understanding it is another matter. While every cell contains the same DNA, a heart cell reads it differently than a brain cell or an immune cell. This makes it incredibly difficult to pinpoint why a small genetic variation, or mutation, might cause a problem in one part of the body but not another. To tackle this, Google's AI division, DeepMind, has introduced AlphaGenome 2. This model is designed to predict how a specific change in DNA affects the behaviour of a particular cell type, a crucial step in tracing the origins of many diseases.
From Broad Tissues to Specific Cells
The original AlphaGenome model, launched in 2025, was a major step forward, helping scientists predict how genetic variants impact tissues on a broad level. However, its successor, AlphaGenome 2, offers a much higher resolution. Instead of looking at tissue averages, it 'zooms in' on individual cells. It was trained on over 107 million cell profiles from 68 different human and mouse tissues, allowing it to learn the subtle differences in how DNA is read from one cell to the next. This conditional model can predict how a DNA change influences a specific cell in a specific state, such as an immune cell that has been activated to fight an infection.
Accelerating Disease Research
The practical implications of this technology are immense. Researchers studying a disease often have a long list of potential genetic culprits to investigate through slow and costly lab experiments. AlphaGenome 2 promises to dramatically speed up this process by helping scientists prioritise which variants are most likely to be disease-causing. In one compelling example, DeepMind reported that AlphaGenome 2 was able to replicate the findings of a study on inflammatory bowel disease in just 30 minutes. The original research, which identified a gene called ETS2 becoming too active in certain gut immune cells, took scientists 18 months of lab work to discover.
The Promise and the Hurdles
The ultimate goal is a future of truly personalised medicine, where treatments can be tailored to an individual’s unique genetic makeup. By understanding the specific cellular mechanisms of a disease, researchers can identify better targets for new drugs. The new AI model is already showing significant improvements in accuracy. In one test on immune cells, AlphaGenome 2 correctly identified 30% of the DNA changes affecting how genes work, nearly double the 16% success rate of its predecessor. However, there are still challenges. The model is a predictive tool, not a diagnostic one, and its findings must be validated in the lab. Furthermore, the increased detail comes at a higher computational cost, and ensuring the ethical use of such powerful genetic data remains a paramount concern for the scientific community.
