Google DeepMind has launched Atlas, an AI-powered platform designed to map every possible DNA change in the human genome. The system provides predictions on how approximately 9 billion potential base substitutions can affect the human body at a molecular level, such as altering the production of specific proteins.
The platform is accessible via a web portal, AI agents called Antigravity, and the AlphaGenome interface. To help researchers prioritize mutations, Google introduced the Variant Impact Score, which uses separate AI models to predict the consequences of DNA changes and rank their molecular effects.
Atlas is built upon AlphaGenome, an AI model released last year, and AlphaMissence, a tool for predicting protein-altering mutations. Unlike previous tools, Atlas extends predictions across the entire genome, including non-coding regions that regulate gene behavior.
The system was trained using public human and mouse genome databases. The resulting dataset is approximately 1 petabyte in size. Google is currently offering non-commercial access to Atlas through its website, with a commercial version planned for release on Google Cloud.