A Bayesian hierarchically structured prior for gene-based association testing with multiple traits in genome-wide...

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Genome-wide association studies (GWAS) often analyze single traits, ignoring correlations between multiple traits for complex diseases.
• Univariate analysis in GWAS may reduce the statistical power to detect genes influencing several related traits.
• Developing methods to analyze multiple correlated traits simultaneously is crucial for comprehensive genetic discovery.

Purpose of the Study:

• To propose a novel multivariate Bayesian model, hierarchically structured variable selection for multivariate data (HSVS-M), for gene-based association testing with multiple correlated traits.
• To develop a method that utilizes only summary statistics from GWAS, accounting for genetic variant and trait correlations.
• To enable estimation of diverse association directions and magnitudes between genes and multiple traits.

Main Methods:

• The multivariate hierarchically structured variable selection (HSVS-M) model, a flexible Bayesian approach, was developed.
• HSVS-M analyzes multiple correlated traits by considering correlations among genetic variants and traits simultaneously using summary statistics.
• The model estimates the directions and magnitudes of associations between genes and multiple traits.

Main Results:

• Simulation studies demonstrated that HSVS-M significantly outperforms competing methods across various scenarios.
• HSVS-M showed particular strength when gene variants associate with traits in similar directions and magnitudes.
• Application to Global Lipids Genetics Consortium GWAS summary statistics identified 15 known risk genes for four lipid traits.

Conclusions:

• The HSVS-M model offers a powerful and flexible approach for gene-based association analysis of multiple correlated traits using GWAS summary statistics.
• This multivariate method enhances the ability to detect genetic associations relevant to complex diseases.
• The successful application to lipid traits highlights the potential of HSVS-M in genetic epidemiology and complex disease research.