BAYESIAN METHODS FOR GENETIC ASSOCIATION ANALYSIS WITH HETEROGENEOUS SUBGROUPS: FROM META-ANALYSES TO...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genetic association studies frequently analyze data from multiple, potentially heterogeneous subgroups.
• Existing statistical tools often struggle to adequately address this heterogeneity, with many meta-analyses assuming homogeneity (fixed effects analysis).
• There is a need for robust statistical methods capable of handling varying degrees of heterogeneity in genetic association studies.

Purpose of the Study:

• To develop and apply novel Bayesian association methods to effectively address heterogeneity in genetic association analyses.
• To provide statistical tools that are easy to apply and can incorporate standard meta-analysis methods as special cases.
• To evaluate these methods using large-scale genetic association studies, including genome-wide association studies (GWAS) and expression quantitative trait loci (eQTL) analyses.

Main Methods:

• Development of Bayesian association methods designed to handle heterogeneity across subgroups.
• The simplest application requires only subgroup point estimates and standard errors for genetic effects.
• These Bayesian methods generalize standard frequentist meta-analysis techniques, including fixed-effects models.

Main Results:

• Application to the Global Lipids consortium GWAS data revealed generally homogeneous genetic effects across studies, contrary to expectations.
• Analysis of cross-population expression quantitative trait loci (eQTLs) demonstrated that eQTLs are typically shared across different continental groups.
• The developed Bayesian methods proved effective in analyzing these large, complex genetic datasets.

Conclusions:

• The novel Bayesian methods provide a flexible and powerful approach for genetic association analyses involving heterogeneous data.
• The findings suggest that genetic effects in lipid-related traits are broadly consistent across diverse study populations.
• Expression quantitative trait loci are largely conserved across human continental groups, with implications for future genetic study design.