Suppose I have a test for determining ‘race’ that is 75% accurate, and suppose I have another test that ascribes ‘disease type’ based on race that is 75% accurate.
We now have tests that are close to 100% accurate, e.g. Am. J. Hum. Genet., 76:268-275, 2005: “Of 3,636 subjects of varying race/ethnicity, only 5 (0.14%) showed genetic cluster membership different from their self-identified race/ethnicity.”
The cluster analysis is rather interpretive, and alternative groupings can be generated. One must ask if the population origins were unknown would the same groupings be generated?
Yes, the results of Rosenberg et al. Science, Vol 298, Issue 5602, 2381-2385 did not use any information about the population affiliation of individuals. Individuals were stripped of all identifying information and were clustered based on allelic values alone. In the resulting clustering the five major continental groups emerged.
Post a Comment