Massively parallel characterization and predictive modelling of neuronal regulatory variation

Area of Science:

• Genomics
• Regulatory Genomics
• Human Genetics

Background:

• Disease-associated genetic variants are often found in noncoding cis-regulatory elements (CREs).
• The functional impact of these variants on gene regulation remains largely unknown.
• Understanding noncoding variant effects is crucial for interpreting genetic risk for diseases.

Purpose of the Study:

• To functionally characterize a large set of genetic variants within candidate CREs.
• To assess the impact of variant frequency on regulatory effects.
• To improve computational models predicting the functional consequences of noncoding variants.

Main Methods:

• Performed a large-scale lentiviral reporter assay in human excitatory neurons (lentiMPRA).
• Quantified regulatory effects of over 46,000 naturally occurring variants across more than 27,000 candidate CREs.
• Analyzed variants near 524 disease-associated genes.

Main Results:

• Developed improved predictive models for regulatory variant effects.
• Found comparable rates of significant allelic effects across common, rare, and singleton variants.
• Demonstrated that variant frequency has limited predictive power for regulatory impact.
• Showed variant effect detectability and magnitude depend on CRE baseline activity and local sequence context.
• Observed regulatory effects distributed across many transcription factors, suggesting combinatorial enhancer function.

Conclusions:

• Established a large-scale functional variant catalog for noncoding regulatory elements.
• Provided a benchmark for developing and evaluating models of noncoding regulatory variation.
• Highlighted the importance of CRE context over variant frequency for regulatory impact.