The problem is not that the kits cannot accurately read off your DNA; they do an admirable job of that. The problem is that our knowledge of the impact of any p

The problem is not that the kits cannot accurately read off your DNA; they do an admirable job of that. The problem is that our knowledge of the impact of any particular allele is exceedingly incomplete. Without a clinical trial validating these tools, most of us in the field find it to be inappropriate to advise someone as to their risk profile on the basis of direct-to-public kits. (I'm not talking about the BRCA tests or other validated, gene-specific tests - I'm talking about 23andMe-style genome-wide kits.) Particularly for complex (polygenic) traits, these throw you into a minefield. Consider two modes of error: (1) The test indicates impending illness. (2) The test indicates no impending illness / good longevity. For (1), imagine that you have a disease-predisposing allele that the kit catches, which is why it thinks you'll be sick. But it doesn't catch (since these kits only target 1/10,000th of the genome) some other allele that protects you from the same disease. So you inappropriately worry about the disease. If disease risk is modifiable by behavior (such as exercise), perhaps it encourages you to work out more often. But, perhaps it will make you feel like you cannot overcome your genetic destiny, so you don't work out at all. The prophecy would then be self-fulfilling. For (2), imagine that the testing kit sees that you have a few protective alleles for disease X - but you also have a novel mutation that is private to your family and not found in the general population. The testing company has no idea that this exists, and it obviously won't be on these testing kits (which overwhelmingly track common, not rare, variants). So now you end up getting disease X a few years down the road, despite all evidence from the kit suggesting that you'd never get it. These are my two favorite failure modes, though there are plenty of others. At the end of the day, although I'm in the field of human genetics, I'm not going to be advising people to get these kits until there are clinical trials validating specific risk predictions. (I'm not saying not to use the kits - but I am saying that it's going to be more for fun than for accurate medical prediction. For now.)