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

Area of Science:

• Genomics
• Developmental Biology
• Computational Biology

Background:

• Understanding noncoding genetic variants' roles in cell-specific gene regulation across human development is challenging.
• Noncoding variants contribute significantly to human traits and diseases, but their functional interpretation remains difficult.

Purpose of the Study:

• To develop a computational framework for predicting the functional impact of noncoding variants in diverse cell types during human development.
• To differentiate the regulatory effects of common and ultra-rare noncoding variants.
• To identify noncoding variants associated with human diseases.

Main Methods:

• Generated over 3 billion predictions of chromatin accessibility using deep learning sequence models across fetal and adult cell types.
• Integrated population genetics data and evolutionary constraint to prioritize functional noncoding variants.
• Developed FLARE (Functional Lasso Analysis of Regulatory Evolution) to identify variants with extreme regulatory effects.

Main Results:

• Common variants exhibit more cell-type-specific regulatory effects, while ultra-rare variants show broader impacts.
• Strongest evidence of purifying selection on variants was observed in fetal neurons.
• FLARE successfully prioritized noncoding variants implicated in childhood disorders, adult brain expression outliers, and schizophrenia heritability.

Conclusions:

• Integrating single-cell chromatin accessibility, population genetics, and deep learning provides a powerful framework for identifying regulatory variants.
• This approach can elucidate the role of noncoding variation in human development and complex diseases.
• The findings highlight the distinct contributions of common and rare variants to regulatory evolution and disease risk.