Case-control association tests correcting for population stratification

Area of Science:

• Population Genetics
• Statistical Genetics
• Genomic Association Studies

Background:

• Unobserved population stratification is a major confounder in case-control association studies, inflating false positive rates.
• Existing methods to account for population structure include Genomic Control (GC) and Structured Association (SA).

Purpose of the Study:

• To extend Structured Association (SA) methods by incorporating phenotypic information for more accurate population structure inference.
• To compare the performance of enhanced SA methods against GC and standard SA in simulations of large case-control studies.

Main Methods:

• Developed extensions to Structured Association (SA) methods, emphasizing the necessity of including phenotypic data.
• Utilized a Wald test statistic over a likelihood ratio test for SA in scenarios with moderate population stratification.
• Conducted simulation studies mimicking realistic large-scale case-control studies with moderate population stratification.

Main Results:

• Incorporating phenotypic information into SA is crucial to prevent systematic bias during population structure inference.
• The Wald test statistic is more appropriate than the likelihood ratio test for SA with moderate population stratification.
• Genomic Control (GC) exhibits high variance in estimating the variance inflation factor and suffers power loss with increased population structure.

Conclusions:

• Structured Association (SA), when correctly implemented with phenotypic data, is superior to Genomic Control (GC) for managing population structure in genetic association studies.
• The proposed SA extensions offer a more robust approach to controlling for confounding in large case-control studies.