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

FLARE deep learning framework predicting noncoding variant effects across cellular and developmental contexts

This Nature Genetics study presents FLARE, a deep learning framework generating 3 billion predictions of chromatin accessibility across diverse fetal and adult cellular contexts. FLARE reveals a dichotomy: common variants are more cell-type-specific, while ultra-rare variants affect more ubiquitous regions. The tool significantly improves prioritization of pathogenic noncoding variants.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

Noncoding variant interpretation remains a major blind spot of diagnostic WGS. FLARE, by integrating developmental cellular context, is a significant methodological advance with potential to improve yields in unsolved genome cases.

Clinical impact: 2/3 · Evidence strength: 3/3 · Novelty: 2/2 · Sample size: 1/1 · Publication status: 1/1 → Total: 9/10