Complete effect-profile assessment in association studies with multiple genetic and multiple environmental factors

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Post-genome-wide association studies (GWAS) research necessitates methods analyzing multiple genetic (Multi-G) and environmental (Multi-E) factors.
• Detecting gene-environment (G×E) interactions is crucial for understanding complex diseases and biological mechanisms.
• Existing methods for Multi-G-Multi-E analyses are often underpowered or ill-suited for complex interaction modeling.

Purpose of the Study:

• To address limitations in current G×E interaction analysis for Multi-G-Multi-E studies.
• To propose a novel kernel machine (KM) approach for constructing G×E kernels and evaluating effect profiles.
• To provide a more interpretable and powerful tool for G×E interaction analysis.

Main Methods:

• Illustrate challenges in constructing kernels for factor-set interactions.
• Propose a new G×E kernel construction method that maintains interpretability.
• Develop KM score tests to assess G, E, and G×E effects individually or combined.

Main Results:

• The proposed KM methods demonstrate superior performance compared to classic and PC regressions.
• Significant power gains were observed, particularly for complex G×E interaction structures.
• The methods show consistent and powerful performance across diverse effect sizes and complexities.

Conclusions:

• The novel KM methods offer a powerful and flexible tool for G×E interaction analysis.
• These methods are suitable for both exploratory and confirmatory research in complex disease genetics.
• The proposed approach enhances the ability to decipher biological mechanisms underlying G×E interactions.