We report here the development of the EMMAX program, taking an expedited mixed linear model approach to correct for sample structure within human GWASs. We demonstrate its effectiveness with the analysis of two human GWAS data sets, including quantitative as well as disease traits. The proposed approach differs substantially from genomic control in that it accounts for inflation owing to population structure in a marker-specific manner, resulting in a modified ranking of association results. Accounting for marker-specific effects can reduce both false positives and false negatives. We discuss this issue in more detail in the Supplementary Note.
