A meta-analytic framework for detection of genetic interactions

Area of Science:

• Population Genetics
• Statistical Genetics
• Genomic Association Studies

Background:

• Substantial heritability remains unexplained by single-nucleotide polymorphism (SNP) genetic variation.
• Detecting genetic interactions is challenging due to the vast number of potential combinations and multiple testing corrections.
• Existing methods may miss higher-order interactions with modest marginal effects.

Purpose of the Study:

• To propose a novel procedure for detecting gene-by-gene interactions.
• To leverage population structure (ancestral differences) among studies to identify interactions.
• To implement this approach within a robust and computationally efficient meta-analytic framework.

Main Methods:

• Developed phylogenY-aware Effect-size Tests for Interactions (YETI).
• Utilized heterogeneity in estimated low-order effect sizes across studies with varying population structures.
• Applied a dimension reduction procedure for scalable searches of higher-order interactions.
• Adapted an existing method assuming strong marginal effects for meta-analysis comparison.

Main Results:

• YETI excels in detecting interactions when studies involve highly differentiated populations, even with small marginal effects.
• The method's advantage is most pronounced under conditions of significant population structure.
• Type-I error and power characteristics were assessed for YETI and a comparative approach.

Conclusions:

• YETI offers a powerful approach for detecting gene-gene interactions by incorporating population structure.
• The meta-analytic framework enhances robustness and computational efficiency, suitable for data-sharing limitations.
• This method advances the extraction of maximal information from genetic association studies.