Improving the power for disease locus detection in affected-sib-pair studies by using two-locus analysis and multiple regression methods

In this paper we present a summary of an analysis of the simulated data (Pr oblem 2) for GAW11. We used sib-pair and affected-sib-pair (ASP) methods to evaluate linkage to the mild form of disease at markers across the genome, in data sets of realistic moderate size (containing between 100 and 300 fa milies selected from the simulated replicates). The true 'answers' were kno wn in advance. Although in most cases we were successful in detecting linka ge to disease in the correct regions, it was often difficult to distinguish these results from false positives elsewhere in the genome. We used two-lo cus methods to see whether the significance was improved by simultaneously modeling linkage to two disease loci, and found a modest increase in signif icance using two-locus methods in several cases. (C) 1999 Wiley-Liss, Inc.