Population substructure and control selection in genome-wide association studies

Area of Science:

• Genetics
• Epidemiology
• Bioinformatics

Background:

• Population stratification (PS) poses challenges in genome-wide association studies (GWAS) by confounding results.
• Effective control selection and PS handling are crucial for accurate genetic association studies.

Purpose of the Study:

• To evaluate the impact of different control selection strategies on PS in GWAS.
• To develop and assess a novel permutation procedure for PS correction in GWAS.

Main Methods:

• Utilized empirical data from two GWAS in European Americans (Cancer Genetic Markers of Susceptibility project).
• Employed principal component analysis (PCA) on 12,898 autosomal SNPs to infer population substructure.
• Developed and tested a novel permutation procedure for PS correction.

Main Results:

• Minor PS confounding observed in original studies (lambda ≈ 1.005-1.025).
• Suboptimal control selection strategies led to larger confounding effects (lambda ≈ 1.062-1.090).
• The novel permutation procedure effectively controlled type I error (lambda ≈ 1.006-1.032), outperforming current methods.

Conclusions:

• Well-designed GWAS, particularly those in European Americans, may have minimal issues with PS.
• Effective PS correction strategies are vital for maintaining statistical accuracy, especially when using less optimal control groups.
• The developed permutation procedure offers improved control of type I error in GWAS analysis.