AlphaGenome 2 maps genetic effects to individual cell types

Google DeepMind’s AlphaGenome 2 predicts how genetic variants may affect specific cell types and states, giving researchers a more detailed way to investigate the biology behind disease. The model is available for non-commercial research through an API, with commercial research limited to a private preview.

Google DeepMind Watch analysis

What happened

AlphaGenome 2 was trained on more than 100 million cell profiles spanning 68 tissues. The Verge reports that Google DeepMind says the model identifies nearly twice as many variants that can change gene expression in cells as its previous state-of-the-art model.

The model can also make predictions for new cell types and biological states without retraining, and may help researchers identify gaps in experimental data. Google DeepMind says researchers can use it through the AlphaGenome API.

Why it matters

A genetic variant’s effect can depend on which cell it appears in and what that cell is doing. More specific predictions could help scientists narrow down which variants and cell types deserve attention in the lab, including when studying possible links to disease.

The finer detail comes at a cost: AlphaGenome 2 is more computationally expensive than its predecessor. That makes it a specialised tool for research, rather than an obvious replacement for broader analyses.

Our read

The useful shift is from asking whether a DNA change matters to asking where and under what cellular conditions it might matter. That could help researchers choose better experiments, though predictions are a guide to investigation, not a diagnosis or treatment. The next test is whether the model’s extra resolution holds up across independent research and practical workloads.

What to watch

• Whether independent researchers reproduce the reported gains. – How researchers use the model to choose and prioritise lab experiments. – Whether its computational cost limits the scale of real-world use.

Discussion spark: Should researchers favour more detailed, computationally expensive predictions when deciding which genetic variants to study, or prioritise tools that can cover more of the genome at lower cost?

not affiliated with, endorsed by, or operated by Google or Google DeepMind