Hierarchical joint analysis of marginal summary statistics-Part I: Multipopulation fine mapping and credible set...

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Population Genetics

Background:

• Genome-wide association studies (GWAS) are advancing through larger sample sizes and a focus on underrepresented populations.
• Multipopulation GWAS leverage diverse linkage disequilibrium (LD) patterns to increase power for detecting risk variants and improve fine-mapping resolution.

Purpose of the Study:

• To expand the single-population Joint Analysis of Marginal SNP Effects (JAM) to a multipopulation framework (mJAM).
• To develop a novel method for constructing credible sets of causal variants in multipopulation GWAS.

Main Methods:

• Implemented a hierarchical model framework for multipopulation analysis (mJAM) that incorporates diverse LD structures.
• Utilized the mJAM likelihood for index variant selection via feature selection approaches (mJAM-SuSiE and mJAM-Forward selection).
• Developed a novel mediation-based approach for constructing credible sets for identified index variants.

Main Results:

• Simulation studies demonstrated mJAM's effectiveness in constructing concise credible sets that include causal variants.
• Real data analysis from a prostate cancer GWAS highlighted practical advantages of mJAM over existing multipopulation methods.
• The mJAM framework successfully identified novel risk variants and improved fine-mapping resolution across diverse populations.

Conclusions:

• The multipopulation Joint Analysis of Marginal SNP Effects (mJAM) framework is a powerful tool for genetic fine-mapping.
• mJAM offers practical advantages and improved performance compared to existing multipopulation GWAS methods.
• This approach enhances the ability to detect and fine-map causal variants by integrating data from diverse populations.