A two-stage mixed-effects model approach for gene-set analyses in candidate gene studies.

In genetic association studies, a gene-set analysis can be more powerful than the separate analyses of multiple genetic variants and can offer unique insights into the genetic basis of many common human diseases. The goal of such an analysis is to study the joint effect of multiple single-nucleotide polymorphisms (SNPs) which belong to certain genes, and these genes are assumed to be involved in a common biological function. Currently, few approaches acknowledge the within-genes and between-genes correlations when testing for gene-set effects. Thus, here we propose a two-stage approach, which in the first stage uses a mixed-effects model with a general random-effects structure to capture the correlation between the SNPs and in the second stage tests for gene-set effects by using the empirical Bayes estimates of the random effects of the first stage as covariates in the model for the longitudinal phenotype. The advantage of this approach is its broad applicability because it can be used for any phenotypic outcome and any genetic model and can be implemented with standard statistical software.