AlphaGenome 2: Google DeepMind Model Maps DNA Effects by Cell Type
Google DeepMind has released AlphaGenome 2, an AI model that predicts how small changes in DNA alter gene activity in specific types of human cells.
Google DeepMind has released AlphaGenome 2, an AI model that predicts how small changes in DNA alter gene activity in specific types of human cells. Scientists hope the finer detail will help trace the roots of disease and point toward treatments.
The company announced the model on November 11, 2026, in a blog post by researchers Adam Gayoso, Natasha Latysheva, Pushmeet Kohli and Žiga Avsec.
Why cell type matters
Nearly every cell in the body carries the same DNA. Yet a heart muscle cell, a skin cell and an immune cell fighting an infection each switch on a different set of genes. Those patterns also shift with what a cell is doing at the time.
That means one genetic variant can behave very differently from cell to cell. A variant may matter only in a specialized cell, or only in a cell activated by disease. In a broad tissue average, its signal can get lost.
What the new model does
AlphaGenome 2 is built to catch those differences. It predicts how variants affect gene regulation in individual cell types and states, rather than across whole tissues.
DeepMind says it trained the model on more than 100 million individual cell profiles. Each prediction is conditioned on a chosen tissue, cell type or cell state. Researchers can also supply conditions drawn from their own single-cell experiments.
Its predecessor, AlphaGenome, was released last year and covered broad tissues. It produced fixed outputs for bulk tissues and cell lines at once, working from a one-million-letter stretch of DNA. The new model evaluates one condition at a time.
How it differs from the atlas
DeepMind also built AlphaGenome Atlas, a platform of predictions for all 9 billion possible single-letter changes in the human genome. AlphaGenome 2 is a separate model release, not an extension of that precomputed atlas.
The work builds on efforts such as the Human Cell Atlas, which is mapping gene expression across millions of cells. DeepMind says predicting how subtle DNA variations change that expression remains an unsolved challenge. It is also a key driver of human traits and disease.
Access and what comes next
AlphaGenome 2 is available for non-commercial research. A private preview is open for commercial research. The announcement describes a tool for scientists studying the cellular drivers of disease, not a clinical product.
