Inferring Gene-by-Environment Interactions with a Bayesian Whole-Genome Regression Model

Area of Science:

• Genetics
• Biostatistics
• Human Health

Background:

• Gene-environment interactions (GxE) are crucial for understanding human traits and diseases, but their contribution is often unclear.
• Large-scale biobanks offer opportunities to study GxE due to extensive genetic and environmental data.

Purpose of the Study:

• To introduce LEMMA (Linear Environment Mixed Model Analysis), a novel Bayesian whole-genome regression model.
• To jointly model main genetic effects and GxE interactions in large datasets.
• To estimate the phenotypic variance attributable to GxE and identify specific genetic loci involved.

Main Methods:

• Developed LEMMA, a Bayesian model for joint genetic and GxE analysis.
• Created an environmental score (ES) representing a linear combination of environmental variables.
• Applied LEMMA to UK Biobank data for BMI, systolic blood pressure, diastolic blood pressure, and pulse pressure.
• Utilized robust standard error estimates to account for heteroskedasticity induced by GxE effects.

Main Results:

• Estimated GxE interactions explain 9.3% (BMI), 3.9% (SBP), 1.6% (DBP), and 12.5% (PP) of phenotypic variance.
• Identified low-frequency genetic variants as major contributors to the observed GxE effects.
• Discovered three loci demonstrating significant interaction with the environmental scores (p<10^-7.3).

Conclusions:

• LEMMA provides a powerful framework for dissecting GxE interactions in large biobanks.
• GxE plays a substantial role in the heritability of common human quantitative traits.
• Future research should leverage LEMMA to further elucidate the genetic architecture of complex diseases.