To classify individuals (from eight diverse ethnic groups), the authors genotyped them for 39 microsatellite markers. They then used a program called STRUCTURE to cluster individuals on the basis of the similarity of their genotypes, and four probable clusters were identified. Some ethnic groups were confined mostly to one cluster, but others fell into more than one. For example, only 24% of the Ethiopian group fell into the cluster that contains Black individuals (the ethnic group to which Ethiopians would normally be assigned), whereas 62% were grouped with Norwegians and Ashkenazi Jews.
Having defined the four genetic clusters, the authors then investigated their biological significance with respect to drug response. They did this by examining the frequencies of 11 functional variants in genes that are involved in drug metabolism. Sure enough, statistically significant differences were found in allele frequencies between some of the clusters. Importantly, some of these differences would have been missed if individuals had been classified according to standard ethnic labels. For example, the frequency of a
DIA4
variant — which leads to loss of protection against the toxic effects of quinones — differed between populations from China and Papua New Guinea, which were assigned to different clusters. Normally, these populations would have been grouped together as Asians, and this important difference would have been missed.
This is a preview of subscription content, access via your institution
