A unified sampling approach for multipoint analysis of qualitative and quantitative traits in sib pairs

Recent advances in molecular biology have enhanced the opportunity to condu ct multipoint mapping for complex diseases. Concurrently, one sees a growin g interest in the use of quantitative traits in linkage studies. Here, we p resent a multipoint sib-pair approach to locate the map position (tau) of a trait locus that controls the observed phenotype (qualitative or quantitat ive), along with a measure of statistical uncertainty. This method builds o n a parametric representation for the expected identical-by-descent statist ic at an arbitrary locus, conditional on an event reflecting the sampling s cheme, such as affected sib pairs, for qualitative traits, or extreme disco rdant (ED) sib pairs, for quantitative traits. Our results suggest that the variance about <(tau)over cap>, the estimator of tau, can be reduced by as much as 60%-70% by reducing the length of intervals between markers by one half. For quantitative traits, we examine the precision gain (measured by the variance reduction in <(tau)over cap>) by genotyping extremely concorda nt (EC) sib pairs and including them along with ED sib pairs in the statist ical analysis. The precision gain depends heavily on the residual correlati on of the quantitative trait for sib pairs but considerably less on the all ele frequency and exact genetic mechanism. Since complex traits involve mul tiple loci and, hence, the residual correlation cannot be ignored, our find ing strongly suggests that one should incorporate EC sib pairs along with E D sib pairs, in both design and analysis. Finally, we empirically establish a simple linear relationship between the magnitude of precision gain and t he ratio of the number of ED pairs to the number of EC pairs. This relation ship allows investigators to address issues of cost effectiveness that are due to the need for phenotyping and genotyping subjects.