Model-free sib-pair linkage analysis: combining full-sib and half-sib pairs

When sampling full-sibs for linkage studies, half-sibs are often available. Not only are half-sibs convenient to sample, but they can sometimes offer greater power than full-sibs. We propose a method to combine the information from full-sibs and half-sibs into a single test for linkage. This method is based on the Haseman and Elston [1972] method of regressing the squared trait-difference for a pair of sibs (either full- or half-sibs) on the estimated proportion of alleles shared identical by descent. To approximate the distribution of the test statistic, we propose a correction factor that considers the correlation among sibs, and demonstrate by simulations that this approximation works well in many situations, although there are some conditions for which the statistic can have an inflated Type-I error rate. The main appeal of our proposed method is the speed at which it can be computed, offering a rapid way to perform genome-wide linkage screens.