My main current interest in personal genomics right now is pure recreation. I don't expect much utility out of it, because a lot of correlations between genes (SNPs, etc. ) and traits/diseases are rather weak. But there are some exceptions. Recently I was temporarily put on a prescription medication and I wanted to check if I was a fast or slow metabolizer. The material you see in the medical literature is that Europeans tend to be slow metabolizers, while Asians tend to be fast metabolizers. Since I'm Asian, I'm probably a fast metabolizer, right? Not so fast! Though I'm geographically Asian (my family hails from Asia), in terms of ancestry South Asians tend to be closer to Europeans, though with some affinity to East Eurasian populations as well. But another issue for me is that I clearly have 10-15% more recent East Asian ancestry, which is not typical in South Asians. In other words, I can't infer with any confidence from generalizations about Asians and Europeans in the American medical literature to my personal status. But that's OK. It turns out that one locus determines most of the effect of this trait, and that locus has been genotyped in 23andMe. Using Promethease I ran my genotype, and it concluded I was a slow metabolizer. This has some utility in terms of when I take my medication. And it's the first time that I can think of 23andMe giving me "actionable" information. More broadly I realized that this sort of genotyping service is particularly useful for those of us who fall between the European/East Asian/African categories used in much of the American medical literature. The main concern is that genetic background might matter. That is, a SNP or set of makers correlated with a disease or trait in one population may not be correlated in another population. But this concern is less of an issue for me now after the past few years. Though some different risk alleles are being found across populations, by and large they're pretty portable. Though the utility of this service for South Asians is obvious, in the American context it will probably be most useful to mixed-race Latinos, who are going to segregate out in traits at large effect loci between their parental populations (if those traits do differ).