ABSTRACT
Introduction Epidemiologists have two m ain approaches to distinguishing genetic and environm ental explanations of ethnic differences in disease risk: studies of m igrants and studies of the relationship of disease risk to proportionate adm ixture in populations of mixed descent. These epidemiological criteria ind icate th a t m ost e th n ic v a ria tio n in d isease risk is a ttr ib u ta b le to environm ent ra th e r th an to genes. For hypertension in West Africans compared with other groups, and for non-insulin-dependent diabetes in many non-European groups compared with Europeans, the epidemiological evidence points strongly to genetic explanations for the ethnic difference in disease risk. This is of p ractical im portance, because, where suitable adm ixed populations exist, it is theoretically possible to map the genes that underlie these ethnic differences in disease risk by a novel approach tha t exploits adm ixture between hum an populations in a m anner analogous to linkage analysis of an experim ental cross. By conditioning on parental ancestry, and combining inform ation from all m arkers on a chromosome in a m ultipoint analysis, it is possible in principle to extract all the inform ation about linkage th a t is generated by adm ixture. M arker sets suitable for genome searches could be assembled by screening m icrosatellites, screening single-nucleotide polymorphisms or using subtractive hybridisation to discover restriction site polymorphisms tha t are informative for ancestry. To combine the m arker data in a m ultipoint analysis and test for linkage, the posterior distribution of ancestry at each locus could be generated by Gibbs sampling and a score test constructed by averaging over this posterior distribution. This adm ixture mapping approach has far greater power than the allele-sharing linkage study designs tha t are now widely used to study the genetics of complex traits in hum ans. Suitable adm ixed populations exist in the Americas, A ustralia, southern Africa and circum polar regions. T here are obvious applications to studying the genetics of hypertension, diabetes, cancer and autoim m une disease.
