Google DeepMind Debuts AlphaGenome 2: AI to Predict DNA's Impact on Human Cells

Google has officially announced AlphaGenome 2, a cutting-edge AI model designed to predict how DNA sequences and their variations influence specific human cell types. By focusing on single-cell biology rather than bulk tissue averages, the model provides scientists with a high-resolution tool to identify cellular causes of diseases and expedite the development of targeted therapies.

Advanced Single-Cell Modeling

Trained on a massive dataset of over 100 million single-cell profiles, AlphaGenome 2 is a conditional model. This architecture allows it to analyze subtle differences within specialized cells, such as immune or cardiac cells, without the need for task-specific retraining. The system has demonstrated the ability to identify nearly twice as many causal DNA variants compared to previous computational methods.

• Mapping long-range interactions between enhancers and target genes.
• Predicting whether a DNA variant increases or suppresses gene activity.
• Reconstructing missing experimental data through cross-modal prediction.

A significant breakthrough of AlphaGenome 2 is its ability to link Genome-Wide Association Studies (GWAS) data to the specific cells where pathologies manifest. Since most genetic variations occur in non-coding regions of the genome, this model provides a crucial link in understanding how these variants actually drive disease processes at a cellular level.

"This turns static cell atlases into a powerful predictive system," the company stated in their announcement.

AlphaGenome 2 is currently available for non-commercial research via the AlphaGenome API. Google also plans to integrate the model as a 'Skill' within its Antigravity agentic development platform, enabling researchers to combine specialized biological modeling with multimodal AI capabilities.