Robust estimation of gene frequency and association parameters

Methods based on generalized estimating equations are proposed for estimation and testing of gene frequency and association parameters using family data. In the problems considered, a marginal model is used to estimate the gene frequency and association parameters specified in the marginal distributions of the individuals. These procedures are robust in the sense that consistent estimates of the parameters of interest and their standard errors are obtained even though the marginal model is not fully correctly specified. In addition, the methods are shown to have good efficiency when compared with maximum likelihood methods. Specific procedures for application of the methods to the cases of allele frequency estimation, testing of linkage and Hardy-Weinberg disequilibrium parameters, and testing for marker-disease association are outlined and illustrated with data examples.