On a randomization procedure in linkage analysis

Although much theoretical work has been undertaken to derive thresholds for statistical significance in genetic linkage studies, real data are often c omplicated by many factors, such as missing individuals or uninformative ma rkers, which make the validity of these theoretical results questionable. M any simulation-based methods have been proposed in the literature to determ ine empirically the statistical significance of the observed test statistic s. However, these methods either are not generally applicable to complex pe digree structures or are too time-consuming. In this article, we propose a computationally efficient simulation procedure that is applicable to arbitr ary pedigree structures. This procedure can be combined with statistical te sts, to assess the statistical significance for genetic linkage between a l ocus and a qualitative or quantitative trait. Furthermore, the genomewide s ignificance level can be appropriately controlled when many linked markers are studied in a genomewide scan. Simulated data and a diabetes data set ar e analyzed to demonstrate the usefulness of this novel simulation method.