Rare variant association testing by adaptive combination of P-values

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Next-generation sequencing (NGS) necessitates advanced statistical methods for detecting rare variants (minor allele frequencies (MAFs) <1%) linked to diseases.
• Individual variant testing is underpowered; group-based association tests pool signals but can be diluted by neutral variants.
• Existing pooling strategies may reduce statistical power due to the inclusion of numerous neutral variants.

Purpose of the Study:

• To develop a novel statistical approach for robust rare variant association testing in NGS data.
• To enhance the power of association tests by adaptively combining signals from multiple variants within a region.
• To address the challenge of neutral variants diluting association signals in pooled analyses.

Main Methods:

• Extension of the [Formula: see text]-MidP method.
• Development of an adaptive combination of P-values approach (ADA) for rare variant association testing.
• Implementation of a truncation threshold on per-site P-values to mitigate noise from neutral variants.

Main Results:

• The proposed ADA method demonstrates superior performance compared to popular burden and non-burden tests.
• ADA effectively identifies disease-associated rare variants even when neutral variants are present in the analysis.
• The adaptive weighting strategy based on MAFs improves the power of association tests.

Conclusions:

• ADA is a powerful and recommended method for analyzing rare variants in NGS data.
• The method effectively handles the high proportion of neutral variants often encountered in functional regions.
• ADA offers improved detection of disease associations compared to existing rare variant testing strategies.