Two-stage global search designs for linkage analysis I: use of the mean statistic for affected sib pairs

Two-stage global search designs for linkage analysis using pairs of affected relatives were shown by Elston et al. [1996] to typically halve the cost of a study compared to a one-stage design. The statistic used for testing linkage in that study was based on the proportion of pairs sharing no marker alleles identical by descent (IBD). However, it has been established that the mean statistic often has the greatest power for full sib pairs [Blackwelder and Elston, 1985; Schaid and Nick, 1990; Knapp et al. 1994]. In this paper, we study optimal two-stage global search designs, in the case of affected full sib pairs, when using the mean test statistic to test for linkage. When dominant genetic variance is present, using the mean statistic is usually more cost efficient than using the proportion of pairs sharing no maker alleles IBD; in the case when there is no dominant genetic variance, the mean statistic leads to a better design, in the sense of being more cost-saving, provided that the relative risk ratio for first-degree relatives is small. The effect of heterogeneity and markers' informativeness is also investigated, the latter using the Linkage Information Content value for sibs.