A unified method for rare variant analysis of gene-environment interactions

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Whole-genome sequencing advances enable rare variant analysis for complex traits.
• Existing methods for gene-environment (GE) interactions, especially for rare variants, are limited.
• Detecting environmental modification of genetic risk is a significant challenge.

Purpose of the Study:

• To develop a novel statistical method for detecting GE interactions involving rare variants.
• To accommodate both binary and continuous traits in related or unrelated individuals.
• To address the challenge of identifying environmental influences on genetic predisposition.

Main Methods:

• Utilized a generalized linear mixed-effects model to incorporate genetic main effects, GE interactions, and sample relatedness as random effects.
• Employed a kernel-based approach for leveraging joint information across multiple rare variants.
• Implemented variance component score tests to enhance computational efficiency.

Main Results:

• Simulation studies confirmed the method's correct type I error rates and adequate statistical power across diverse scenarios.
• The method demonstrated robustness with varying genetic effects, interaction directions, and proportions of causal variants.
• Application to the Framingham Heart Study successfully identified a gene-environment interaction related to smoking and body mass index.

Conclusions:

• The developed set-based GE test is computationally efficient and versatile for binary and continuous phenotypes.
• The method effectively accounts for complex sample relatedness, including familial or cryptic relationships.
• This approach advances the ability to detect gene-environment interactions, crucial for understanding complex trait etiology.