DeepMind launches AlphaGenome 2, a finer‑grained AI for cell‑level genetics
What happened
DeepMind researchers announced AlphaGenome 2, the second generation of their genome‑predicting artificial intelligence. The model builds on the original AlphaGenome released last year and is designed to predict the impact of genetic variants on specific cell types and biological states. It was trained on more than 100 million single‑cell profiles drawn from 68 human tissues, giving it a detailed view of how the same DNA mutation can behave differently in heart cells, skin cells, immune cells and many others.
According to DeepMind, AlphaGenome 2 can identify almost twice as many variants that alter gene expression in cells compared with its predecessor. The system can also make predictions for newly observed cell types without additional training, and it can highlight gaps in existing experimental data, suggesting where laboratory work might be most valuable. The researchers describe the model as turning static cell atlases into a “predictive engine for human genetics,” linking population‑scale genetic findings to cellular mechanisms that may drive disease.
The company is releasing AlphaGenome 2 through a private preview of its AlphaGenome API for non‑commercial research, with limited commercial access planned. While the new model offers greater specificity, it also requires more computational resources, making it less suitable for large‑scale genome‑wide atlases at present. DeepMind plans to expand the training data, including under‑represented cell states and potentially data from other species, with a long‑term vision of supporting virtual‑cell simulations.
Why it matters
Understanding how genetic variants influence individual cell types is a critical step for biomedical research. In many diseases, the same mutation can have harmless effects in one tissue while causing pathology in another. By providing cell‑level predictions, AlphaGenome 2 gives scientists a tool to prioritize which variants merit deeper investigation, potentially accelerating the discovery of disease mechanisms and therapeutic targets.
For businesses and research teams in Rwanda and across East Africa, the model could help local genomics projects make more informed choices about which genetic markers to study in the context of regional health challenges. Access to a predictive engine that highlights data gaps may guide laboratory investments, while the ability to simulate cellular responses could inform early‑stage drug discovery collaborations with global partners.
Key takeaways
- AlphaGenome 2 predicts DNA variant effects at the level of individual cell types and states.
- The model is trained on over 100 million single‑cell profiles from 68 human tissues.
- It is now available for non‑commercial research via a private API preview.
AI-assisted summary published by the Diolichat Team. See our editorial policy. Spotted a mistake? Email info@diolichat.co.rw.