Originally posted by twhitehead
It is almost certainly possible already to search for genes that correlate with certain desirable characteristics including health and intelligence. The danger of putting multiple such genes together is that they may not work as planned. Two genes that individually result in greater intelligence may when put together result in mental problems or even some totally unrelated unforeseen issue.
I agree with this. But there are still several big issues with these correlations:
1) The biggest one for me is there is no way to show that gene variants are actually
causal to the phenotype being studied. Genetic information is often linked to other genes, so until you do the experiment (in humans) you don't really know where the "effect" comes from. Many of the strong correlations in gene variants can also break down along race/ethnic lines. Since these populations share lots of other common gene variants, it blurs the significance of one DNA polymorphism.
2) They typically are very low-confidence predictors. So, if you have one genetic variant you are only at a slightly greater risk for such and such disease.
And 3) is to your point: off-target effects. Not only can two disease-protective genes have interrelated (and possibly oppositional) functions, they also have multiple functions individually. As quoted in the study linked below, one variant "decreases the risk of Crohn’s disease but increases the risk of rheumatoid arthritis and type 1 diabetes."
Anyway, this is a good article that gets at the high complexity of the issue. It may be solvable but, even at the single-gene/single-phenotype level, it's a long way from being well-understood. http://www.nature.com/nrg/journal/v14/n7/pdf/nrg3461.pdf