MR-MEGA

MR-MEGA performs trans-ethnic meta-regression to detect and fine-map complex trait association signals across diverse populations by modeling ancestry-correlated heterogeneity in allelic effects.

Key Features:

• Trans-ethnic meta-regression: Models allelic effects as functions of genetic variation axes derived from a matrix of mean pairwise allele frequency differences between GWAS conducted in different populations.
• Quantification of ancestry-correlated heterogeneity: Estimates and tests the extent of heterogeneity in allelic effects that is correlated with ancestry.
• Increased power and fine-mapping resolution: Demonstrated by extensive simulations to have greater power to detect associations and improved fine-mapping resolution, particularly at loci with a single causal variant, compared to traditional fixed- and random-effects meta-analysis.
• Computational efficiency: Provides performance comparable to MANTRA while reducing computational cost for large-scale genetic studies.

Scientific Applications:

• Detection of complex trait loci: Effective for identifying loci when causal variants are shared across ancestry groups and when allelic-effect heterogeneity correlates with ancestry.
• Fine-mapping of genetic variants: Applied to type 2 diabetes susceptibility loci (including CDKAL1) to obtain credible sets containing six or fewer variants for multiple association signals.
• Trans-ethnic GWAS of kidney function: Applied to a cohort of 71,461 individuals, revealing stronger association signals than fixed-effects meta-analysis when heterogeneity correlated with ancestry.

Methodology:

Implements trans-ethnic meta-regression that models allelic effects as functions of genetic variation axes derived from a matrix of mean pairwise allele frequency differences between GWAS in different populations, with performance assessed via extensive simulations and comparisons to fixed-effects, random-effects, and MANTRA approaches.