Estimation of allele frequencies with data on sibships

Allele frequencies are generally estimated with data on a set of unrelated individuals. In genetic studies of late-onset diseases, the founding indivi duals in pedigrees are often not available, and so one is confronted with t he problem of estimating allele frequencies with data on related individual s. We focus on sibpairs and sibships, and compare the efficiency of four me thods for estimating allele frequencies in this situation: (1) use the data for one individual from each sibship; (2) use the data for all individuals , ignoring their relationships; (3) use the data for all individuals, takin g proper account of their relationships, considering a single marker at a t ime; and (4) use the data for all individuals, taking proper account of the ir relationships, considering a set of linked markers simultaneously. We de rived the variance of estimator 2, and showed that the estimator is unbiase d and provides substantial improvement over method 1. We used computer simu lation to study the performance of methods 3 and 4, and showed that method 3 provides some improvement over method 2, while method 4 improves little o n method 3. (C) 2001 Wiley-Liss. Inc.