This project is concerned with statistical genetics problems, particularly in the area of developing and evaluating methods for mapping human trait variation. We have been investigating multi-locus approaches that provide increased efficiency across models of phenotype, and robustness with respect to the random mating assumption. We have developed a novel approach to statistical mapping of genetic variants predictive of traits such as disease susceptibility and adverse reactions to drugs. The approach is based on an observation that in a region of genetic association, the extent of correlation between alleles (linkage disequilibrium, LD) can be different between the case and the control groups. The extent and the pattern of LD can be compared between samples by a visual inspection of LD """"""""color plots"""""""". However, patterns conveyed by such plots are subject to random variation, and a statistical assessment of the LD difference is necessary. A difficulty in constructing an """"""""LD contrast"""""""" test is that the deviations from the Hardy-Weinberg (random mating) proportions are likely to manifest in phenotype-based samples in the presence of genetic association. The deviations from random pairing of haplotypes are likely to occur even if the underlying population is in Hardy-Weinberg equilibrium. Thus, standard techniques of phasing haplotypes are unreliable. The LD contrast method deals with this issue. The method is found especially efficient when high susceptibilities are linked to a pair of """"""""yin yang haplotypes"""""""", i.e. haplotypes carrying different alleles at many SNP sites. ? ? Dr. Kyoko Shibata (Research Fellow) has been working in the direction of improving the likelihood-based statistical methods for association of haplotypes and diplotypes with continuous traits. Dr. Shibata has been applying her methods to study genetic components of pain sensitivity.
