can be used to detect haplotypic effects, it is also able to capture epistatic effects due to interactions between two haplotypes in a diplotype. We conducted two types of haplotype association analysis. ‘Overall’ CHM tests in sliding windows of varying size provide a single P-value for a set of SNPs in a given window. Next, given the overall significance, tests for individual haplotypes were conducted to determine specific haplotypes to which the overall association was due. The individual tests were carried out using the CHM as well as a likelihood-based haplotype association method by Shibata et al. (54,55). To make a comparison of the overall CHM tests with the HTR, we replicated the analysis using the HTR. The CHM appeared to have provided greater evidence of association, which is possibly due to non-additive epistatic interactions (data not shown).
