Approaches to detect genetic effects that differ between two strata in genome-wide meta-analyses: Recommendations...

Area of Science:

• Human Genetics
• Statistical Genomics
• Population Genetics

Background:

• Genome-wide association meta-analyses (GWAMAs) can reveal genetic effect differences between population strata (e.g., sex differences in body fat distribution).
• Existing methods for identifying these between-strata differences vary, leading to uncertainty about the most effective approach.

Purpose of the Study:

• To compare various statistical approaches for identifying between-strata differences in genetic effects using stratified GWAMA results.
• To evaluate the performance (type I error and power) of these approaches under different strata designs and difference types.

Main Methods:

• Simulations and analytical comparisons were used to evaluate different methods for detecting between-strata genetic effect differences.
• Scenarios included equal and unequal strata sizes, and different types of between-strata differences (opposite directions, predominant effects).
• Real data from the GIANT consortium (>175,000 individuals) was used to exemplify the impact of these approaches on detecting sex differences in body fat distribution.

Main Results:

• For equal-sized strata, a genome-wide difference test without filtering is best for opposite-direction effects, while filtering followed by a difference test is optimal for predominant effects.
• A combination of both approaches is recommended when the type of difference is unknown.
• For unequal strata, the best method depends on whether the effect is predominant in the larger or smaller stratum; some methods violate type I error control.

Conclusions:

• The study provides evidence-based guidelines for selecting the most appropriate method to detect between-strata genetic differences in future GWAMAs.
• Identifying stratum-specific genetic effects enhances understanding of underlying biological mechanisms, such as sex differences in complex traits.