DeepMind releases AlphaGenome 2 for single-cell variant prediction
Original title: Google Deepmind shifts genomics AI from bulk tissue predictions to single cell type resolution with AlphaGenome 2
The conditional model is available for noncommercial research by API and is not approved for clinical diagnostics.
Google DeepMind released AlphaGenome 2, a genomics model that predicts how DNA variants affect a specified human cell type rather than bulk tissue. DeepMind says it was trained on more than 107 million cell profiles from 68 tissues and can also infer chromatin accessibility and DNA methylation. On the TenK10K immune-cell benchmark it identified about 30 percent of causal variants, versus 16 percent for a fine-tuned AlphaGenome 1. Commercial access is a closed preview, and DeepMind plans later integration with Google Antigravity and Gemini 4.
Unconfirmed: Benchmark and accuracy figures are Google DeepMind’s own claims.
- It takes a DNA sequence plus a biological state and predicts effects for that cell type.
- Direction of gene regulation was correct in 43 percent of TenK10K cases, versus 23 percent previously.
- Precomputed predictions cover hundreds of cell and tissue conditions for GWAS analysis.
- Researchers can condition it on gene-activity data from a few dozen previously unseen cells.
- Vendors
- Organizations
- Google DeepMind
- AI models
- AlphaGenome 2AlphaGenome
From the original article
4 min read at the-decoder.com
Google Deepmind has unveiled AlphaGenome 2, an AI model that analyzes genetic variants for individual cell types rather than across bulk tissue. Trained on more than 107 million cell profiles, it identifies nearly twice as many disease-relevant gene changes as its predecessor. The model is available for noncommercial research through an API and is set to be integrated into Google's agentic platform Antigravity.
By Alan Truly · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.