Scalable and accurate rare variant meta-analysis with Meta-SAIGE

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Meta-analysis increases power for rare variant association tests by combining cohort data.
• Current methods struggle with type I error control for low-prevalence traits and are computationally demanding.

Purpose of the Study:

• Introduce Meta-SAIGE, a scalable method for rare variant meta-analysis.
• Improve type I error control and computational efficiency in phenome-wide analyses.

Main Methods:

• Meta-SAIGE accurately estimates the null distribution to control type I error.
• Reuses linkage disequilibrium matrices across phenotypes for computational efficiency.
• Validated using UK Biobank whole-exome sequencing data.

Main Results:

• Meta-SAIGE effectively controls type I error and matches the power of pooled individual-level analysis.
• Identified 237 gene-trait associations across 83 low-prevalence phenotypes using UK Biobank and All of Us data.
• 80 associations were significant in the meta-analysis but not in individual datasets, highlighting enhanced discovery power.

Conclusions:

• Meta-SAIGE offers a powerful and computationally efficient approach for rare variant meta-analysis.
• The method is particularly effective for low-prevalence binary traits.
• Meta-SAIGE significantly enhances the discovery of gene-trait associations.