Google DeepMind launches AlphaGenome to analyze DNA impact
Google DeepMind just dropped AlphaGenome, an AI model built to predict how tiny changes in your DNA might affect gene activity and health.
It can scan huge stretches of genetic code—up to a million base pairs—and works with both human and mouse data.
The goal? To help researchers figure out how genes work and what goes wrong in diseases, by looking at both the obvious and hidden parts of our DNA.
How to access the model
AlphaGenome offers super-detailed predictions for over 5,000 human and 1,000 mouse genomic features—like where genes start or stop, RNA output, and splicing—all across different cell types.
It combines 11 types of predictions into one tool, making life easier for scientists.
You can access it via API for non-commercial research (no word yet on commercial pricing).
Why AlphaGenome is better than older models
Unlike older models like DeepSEA or Basenji that only analyze short bits or fewer features, AlphaGenome covers way more ground—including the 98% of our genome that controls gene activity but doesn't make proteins.
This gives researchers a much clearer picture when studying genetic variants linked to disease.