CMAX3: A Robust Statistical Test for Genetic Association Accounting for Covariates

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Genome-wide association studies (GWASs) commonly use logistic regression with additive genetic models for binary outcomes.
• Mis-specifying genetic models in GWASs for complex diseases can lead to significant power loss.
• The MAX3 test offers robustness across unknown genetic models but lacks covariate adjustment capabilities.

Purpose of the Study:

• To extend the MAX3 test for robust genetic association analysis incorporating covariate adjustment.
• To develop a covariate-adjusted MAX3 (CMAX3) test within a logistic regression framework.
• To provide an asymptotic formula for p-value calculation and demonstrate its application.

Main Methods:

• Developed the covariate-adjusted MAX3 (CMAX3) test based on logistic regression.
• Utilized a likelihood framework to enable adjustment for covariates such as age and gender.
• Derived an asymptotic formula for calculating the p-value of the CMAX3 test.

Main Results:

• The CMAX3 test demonstrated robust efficiency comparable to the original MAX3.
• The test effectively adjusted for covariates, enhancing its applicability in GWASs.
• Simulation studies confirmed desirable performance under null and alternative hypotheses.

Conclusions:

• The CMAX3 test provides a powerful and robust tool for GWASs, accommodating unknown genetic models and covariates.
• This method addresses limitations of previous approaches, increasing power and applicability in complex disease genetics.
• The developed R code facilitates the implementation and application of the CMAX3 test in genetic research.