Figure 1.

Different LD patterns can yield different patterns of association. Hypothetical haplotypes in an associated region and their effects on disease risk are shown for a European-derived population (A) and an African-derived population (B). In European-derived populations, several SNPs show equivalent signals of association, including the causal SNP (marked by jagged lines). Two of these are in HapMap, and have been tested via genotyping or imputation, permitting the effect of the causal SNP (which is not in HapMap) to be detected indirectly. In African-derived populations, the causal SNP is rarer and is no longer strongly correlated with the surrounding SNPs in HapMap, so the surrounding SNPs will not show strong association. Thus, a fine-mapping approach based only on HapMap SNPs but without additional resequencing may fail to detect a signal in the African-derived population.