A GENERAL STATISTICAL-MODEL FOR DETECTING COMPLEX-TRAIT LOCI BY USINGAFFECTED RELATIVE PAIRS IN A GENOME SEARCH

Scanning of the human genome by use of affected relative pairs and den se sets of highly polymorphic markers or by emerging techniques such a s genomic mismatch scanning (CMS) is making it possible to identify th e genetic etiology of a disease through detection of susceptibility lo ci. We present a general statistical model and test to detect disease genes, using affected relative pairs and either markers or GMS technol ogies in a genome search. There are an exact test and large-sample nor mal approximation that control for the elevated probability of false d etection of linkage in a genome search. The approach can be used to de termine the sample size needed to obtain a prespecified power to detec t a disease gene in the presence of etiologic heterogeneity for a sing le class or mixture of relative classes, with any number of markers or clones, marker PIC values, or mapping function. The approach is used to examine differences in performance of markers and GMS technologies in a common statistical framework and to provide practical information for designing studies of complex traits.