Researchers introduce Seal AI model for gene regulation in developing human brain
A team led by Y. Hao and O. G. Troyanskaya published Seal, a transfer-learning framework that predicts gene expression and variant effects across 26 brain regions, 30 cell types, and 7 developmental stages.
Key points
- Seal models gene expression across 26 regions, 30 cell types, 7 developmental stages
- Finds de novo regulatory variant burden in fetal excitatory neurons in autism cohort
- Model code and data publicly available on GitHub and Zenodo
Seal links genetic variants to transcriptional regulators in a spatiotemporal context. Applied to GWAS data, it identifies shared and distinct regulatory architectures across six neuropsychiatric conditions. In the Simons Simplex Collection autism cohort, Seal finds a significant burden of de novo regulatory variants in transient fetal excitatory neurons, revealing cell-type- and stage-specific effects. Benchmarks show transfer learning significantly lowers mean squared error (P = 0.01) and improves predictions for neurodevelopmental genes (P = 1.1×10⁻¹⁸). The authors declare no competing interests; funding came from the Simons Foundation and Flatiron Institute.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Anthropic reports small AI models may self-replicate on IoT devices within months · 27 src
- MIT study finds AI solved 35% of theory problems by mid-2026 · 2 src
- Microsoft Research introduces Probabilistic Bias Correction for longer-term weather forecasts · 3 src
- TIER IV funds Autoware research center at Technical University of Munich · 1 src
- MIT professor Eric So warns of AI cognitive atrophy · 1 src
Comments
via GitHub Discussions