Correcting for measurement error in individual ancestry estimates in structured association tests

Area of Science:

• Population Genetics
• Statistical Genetics
• Genomic Association Studies

Background:

• Estimates of individual admixture proportions using ancestry informative markers (AIMs) are crucial for genetic studies of admixed populations.
• These estimates are inherently error-contaminated, reflecting a measurement of underlying true ancestry proportions.
• Population stratification can inflate Type I errors and reduce power in genetic association tests.

Purpose of the Study:

• To theoretically explain and simulate the error-contaminated nature of AIM-based admixture proportion estimates.
• To evaluate the utility of measurement error correction (MEC) methods for controlling confounding effects of admixture in genetic association tests.
• To identify the most effective MEC method for admixed populations.

Main Methods:

• Theoretical derivations of measurement error models for admixture proportions.
• Development of methods to estimate an upper bound for measurement error variance.
• Simulation studies comparing four MEC methods, including quadratic MEC (QMEC), under a two-population admixture model.

Main Results:

• AIM-based admixture proportion estimates are shown to be error-contaminated measurements.
• Inclusion of these estimates as covariates can bias regression models and impair control of admixture confounding.
• The QMEC method effectively controlled Type I error rates to their nominal level, outperforming other tested MEC methods.

Conclusions:

• Measurement error correction is essential for accurate genetic association testing in admixed populations.
• The QMEC method provides a robust approach to address admixture confounding.
• Accurate control of population stratification using QMEC preserves the integrity of genetic association study findings.