Incorporating model uncertainty in detecting rare variants: the Bayesian risk index

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Rare genetic variants with low minor allele frequencies pose challenges for traditional association studies.
• Existing methods often lack the power to detect associations involving multiple rare variants within a genomic region.
• Common genetic markers may not capture the full spectrum of rare variant associations with disease risk.

Purpose of the Study:

• To develop and validate a novel analytical framework for investigating the aggregate and specific effects of multiple rare variants within genomic regions.
• To enhance the statistical power for detecting associations between rare variants and disease risk.
• To identify specific rare variants contributing to disease risk within associated regions.

Main Methods:

• A Bayesian approach is employed to construct a risk index based on multiple rare variants.
• Model uncertainty is incorporated for variant selection and effect direction.
• The framework allows for both group-level and variant-specific inference.

Main Results:

• Simulations demonstrate that the proposed methodology exhibits increased power compared to popular rare variant analysis methods for detecting global associations.
• The approach successfully identifies specific variants driving regional associations.
• The method was applied to sequence data from the WECARE Study of second primary breast cancers.

Conclusions:

• The novel Bayesian framework offers enhanced power for detecting rare variant associations with disease risk.
• This approach provides a robust method for identifying specific genetic variants contributing to disease.
• The methodology is applicable to real-world genetic sequence data, advancing the study of complex diseases.