Decoding common and rare noncoding variant effects across cellular and developmental contexts.

Deep learning predictions reveal that common noncoding variants are cell-type-specific, while ultra-rare variants have broader, more impactful effects across human development.

• Why it matters: Understanding how noncoding variants influence specific cell types during development is crucial for deciphering genetic contributions to human health and disease, yet remains a major challenge.
• What they did: The study generated 3 billion predictions from deep learning models of chromatin accessibility across diverse fetal and adult cell types, and developed FLARE to prioritize variants based on evolutionary constraint.
• The result: Findings show ultra-rare variants exert widespread effects, especially in fetal neurons, and FLARE effectively identifies variants linked to childhood disorders, adult brain expression outliers, and schizophrenia heritability, advancing regulatory genomics.