The development

What happened

Google DeepMind has introduced AlphaGenome 2, an AI model that predicts how DNA may affect gene regulation in a specified cell type or biological state. It is available for non-commercial research through the AlphaGenome API and Colab tutorials; commercial access is a private preview. DeepMind explicitly says the model is not validated or approved for clinical or diagnostic use.

Key facts

  • The model takes a DNA sequence of up to one million base pairs together with one biological condition at a time, such as a tissue, cell type or cell state. This condition-specific approach differs from the original AlphaGenome model's parallel predictions of bulk measurements.

  • DeepMind says training combined bulk and single-cell human and mouse data, including more than 107 million cell profiles spanning 68 tissues. The model predicts gene expression, chromatin features and histone marks, and adds DNA methylation predictions.

  • According to the announcement, researchers can condition predictions on new cell states using RNA-sequencing profiles without retraining the model. That is a reported modeling capability, not proof that a predicted regulatory effect is biologically causal.

  • The 11 November release provides non-commercial research access and tutorials. Commercial access is limited to a private preview; proposed Antigravity agent skills are described as future work rather than an available feature.

The context

Why it matters

Gene regulation depends on biological context, so cell-specific predictions could help researchers prioritize hypotheses about non-coding DNA variants. The announcement does not make the model a diagnostic tool, establish clinical benefit or replace experimental validation. This article reports the developer's release claims, not an independent model evaluation.