Bias correction with a single null marker for population stratification in candidate gene association studies

Area of Science:

• Genetics
• Population Genetics
• Statistical Genetics

Background:

• Population stratification, a confounder by ethnicity, introduces bias in genetic association studies.
• Unlinked genetic markers can adjust test statistics but their role in correcting effect estimate bias is unclear.

Purpose of the Study:

• To evaluate the bias correction potential of a single null marker (M) in candidate gene (G) studies.
• To assess methods for correcting biased effect estimates caused by population stratification.

Main Methods:

• Logistic regression modeling controlling for a null marker (M).
• Subtracting the regression coefficient of M from the coefficient of G.
• Analysis considering marker distributions across ethnicities and baseline disease risks.

Main Results:

• Controlling for M in logistic regression significantly reduced odds ratio biases when M distribution varied across ethnicities.
• Subtracting M's coefficient from G's coefficient further reduced or eliminated bias when M and G distributions were similar.
• Bias correction effectiveness depended on marker distributions, ethnic disease risk differences, and gene's effect.

Conclusions:

• A single null marker can effectively reduce bias from population stratification in genetic association studies.
• The choice of marker and analytical approach significantly impacts bias correction.
• Careful consideration of marker distributions and statistical methods is crucial for accurate genetic association studies.