Mixed Linear Model Approaches of Association Mapping for Complex Traits Based on Omics Variants

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Precise prediction of complex traits is hindered by incomplete understanding of genetic effects, particularly gene-by-gene (GxG) and gene-by-environment (GxE) interactions.
• Advances in high-throughput technologies have generated large-scale omics data (genomics, transcriptomics, proteomics, metabolomics), but their analysis, especially for interactions, is computationally intensive.
• Integrating diverse omics data and environmental factors into analyses presents significant challenges.

Purpose of the Study:

• To develop and validate efficient computational approaches for dissecting complex genetic architectures.
• To simultaneously estimate various genetic effects, including main effects, GxG epistasis, and GxE interactions, from large-scale omics data.
• To assess the heritability of specific genetic effects within complex traits.

Main Methods:

• Proposed mixed linear model approaches enhanced with Graphic Processing Unit (GPU) computation.
• Simultaneous estimation of genetic main effects, GxG epistasis effects, and GxE environment interaction effects.
• Application to large-scale omics data for complex traits and validation through mouse data analysis and Monte Carlo simulations.

Main Results:

• Demonstrated unbiased estimation of genetic effects and environment interaction effects.
• Achieved high statistical power in estimating these complex genetic interactions.
• Successfully applied the methods to mouse data, confirming their practical utility.

Conclusions:

• The proposed GPU-accelerated mixed linear model approaches provide a powerful and efficient solution for analyzing complex genetic architectures.
• These methods enable accurate estimation of GxG and GxE interactions, crucial for understanding complex traits.
• The approach facilitates robust heritability estimation for specific genetic effects in large omics datasets.