Effect of genetic architecture on the power of human linkage studies to resolve the contribution of quantitative trait loci

Abstract

The effect of genetic architecture (linkage relationships, dominance and two forms of non-allelic interaction) on the power of marker studies to detect, locate and analyse the contributions of specific quantitative trait loci (QTLs) to continuous human traits is considered for randomly mating populations in linkage equilibrium under a two-locus model. The expected regression of the within-sibling-pair mean-square on number of alleles identical by descent (IBD) at two marker loci is explored for every possible pair of markers over a region of the genome containing two QTLs linked loosely (50 CM) or more tightly (20 CM). For the cases examined, it is shown that epistasis between the pair of QTLs reduces considerably the total amount of information available for the location and analysis of the QTL effects. The overall effects of epistasis are more marked when there are duplicate gene interactions (i.e. genes operate in parallel) than when there are complementary interactions (i.e. genes operate in series). However, when there are complementary interactions, the regression approach is almost certain to fail to detect any evidence of epistasis. The numerical analysis suggests that methods of QTL analysis based on IBD in humans are unlikely to offer the resolving power that is desirable if QTLs are to be located precisely unless inheritance is very simple or prohibitively large numbers of highly selected individuals are available.