Pairwise linkage disequilibrium under disease models.

Many genetic studies of disease association rely heavily on linkage disequilibrium (LD) patterns between pairs of markers to detect susceptibility markers. This is true of large-scale positional mapping approaches as well as haplotype construction, selection of tagging single-nucleotide polymorphisms and population genetic analyses. Whereas the distribution of different LD measures has been investigated for randomly selected chromosomes from populations undergoing a variety of demographic effects, little is known about LD within disease-affected samples, and how various disease models influence the difference in LD between patients and the general population. As whole-genome efforts are now underway to characterize and utilize LD patterns in randomly sampled individuals, knowledge about the extent that LD differs between patients and the general population becomes crucial. Such information will allow investigators to design improved mapping experiments and better understand haplotype information arising from such experiments. In this paper, we explore two-site LD measures in the context of single gene disease models. Analytic expressions are presented for infinite populations and properties of sampling densities are reported for different disease models. Interestingly, results indicate that 'underdominant', some dominant, recessive and 'protective' disease models generate weaker LD levels in patients compared to the general population, whereas other models produce stronger LD among affected individuals. Analytic results are also presented for the ratio of LD in patients to the LD in the general population as a function of recombination fraction using a Haldane model. In addition, we explore the impact of various allele frequency combinations on LD differences.