Leveraging longitudinal data to boost statistical power for gene-environment interaction analysis

Area of Science:

• Genetics
• Bioinformatics
• Statistical genomics

Background:

• Gene-environment interaction (G×E) analyses are vital for genetic discovery, heritability, and precision medicine.
• Current G×E methods primarily use cross-sectional data, limiting the use of valuable longitudinal data.
• There is a need for efficient G×E methods that can leverage longitudinal data in large-scale studies.

Purpose of the Study:

• To introduce SAGELD, a scalable and accurate genome-wide G×E method designed for longitudinal traits.
• To enable G×E analyses in large datasets while controlling for sample relatedness.
• To demonstrate the advantages of using longitudinal data for G×E studies.

Main Methods:

• SAGELD employs matrix projection for test statistics construction.
• The SPA_GRM framework is utilized for efficient control of sample relatedness.
• The method was validated through simulations and UK Biobank data analysis.

Main Results:

• SAGELD achieves significant speedups (10- to 10,000-fold) compared to existing methods.
• The method demonstrates greater statistical power than cross-sectional G×E analyses.
• Identified 74 loci for genetic × age interactions and 5 loci for genetic × adiposity interactions using UK Biobank data.

Conclusions:

• SAGELD offers a scalable and powerful approach for genome-wide G×E analyses with longitudinal data.
• Leveraging longitudinal data in G×E analyses significantly enhances genetic discovery.
• The findings support the utility of SAGELD for advancing precision medicine through comprehensive genetic insights.