An efficient multi-locus mixed-model approach for genome-wide association studies in structured populations

Area of Science:

• Population genetics
• Statistical genomics
• Bioinformatics

Background:

• Population structure introduces genome-wide linkage disequilibrium, causing confounding in genome-wide association studies (GWAS).
• Existing mixed models effectively address small-effect loci but may not fully account for loci with larger effects.

Purpose of the Study:

• To propose a general multi-locus mixed model (MLMM) for mapping complex traits in structured populations.
• To enhance the power and control the false discovery rate (FDR) of GWAS in the presence of population structure.

Main Methods:

• Developed a novel multi-locus mixed model (MLMM) to simultaneously account for population structure and multiple genetic loci.
• Utilized simulations to compare the performance of MLMM against existing methods.
• Applied the MLMM to human and Arabidopsis thaliana datasets, integrating prior linkage mapping information via a Bayesian approach.

Main Results:

• Simulations demonstrated that MLMM significantly outperforms existing methods in both statistical power and false discovery rate control.
• Application to real data identified novel trait associations and provided evidence for allelic heterogeneity in humans and A. thaliana.
• The MLMM approach is computationally efficient, enabling the analysis of large-scale genomic datasets (n > 10,000).

Conclusions:

• The proposed multi-locus mixed model is a robust and efficient method for complex trait mapping in structured populations.
• MLMM offers improved accuracy and power for genome-wide association studies, facilitating the discovery of genetic variants underlying complex traits.
• The method's flexibility allows for the integration of prior biological knowledge, further enhancing its utility in genetic research.