Earlier this week I sketched out the general theoretical basis for not denying unexpected deviations from expectation, so to speak, when it comes to quantitatve traits. The main issue is that varying genetic backgrounds leave unaccounted for gene-gene interactions, and so our predictions when two populations are crossed maybe confounded (within a population ceteris paribus is far more likely to hold). In any case, I thought I'd give you two obvious examples from humans. First, in 2005 Helgadottir et. al. found that African Americans are at greater risk for myocardial infarction vis-a-vis their parental populatons, Africans and Europeans, because of a combination of alleles of one population against the genetic background of the other (genetic and historial studies tend to converge upon a median admixture proportion of 20-25% European and 75-80% African in black Americans, with variance of course between subpopulations and families). Since genomic data suggest that the Out of Africa event and expansion into Eurasia induced multiple selective sweeps in populations which left the ur-heimat I would not be surprised if more studies like this emerge which suggest decreases in fitness because of problematic genetic combinations. As I've said before, racial admixture increases variation and genetic diversity, and I see no reason why this would not result in an increase in the proportion of those who are far more and less fit than is in the norm in major racial groups (corrected for some possible masking of deleterious recessives and so hybrid vigor). The Neandertal-modern introgression story was in part an illustration of how novel genetic combinations may have unexpected positive benefits. Though on average I think racial admixture is probably a wash for most populations I do believe that the likelihood for the arrival of a genuine Übermensch will increase as powerful assortative mating across a few valued characters proceeds apace. A second example of epistasis is the Pakistani family that can't feel pain. Note:
The SCN9A gene is active both in nerves that mediate pain and in those of the sympathetic nervous system, which controls vital bodily functions like heart rate. But for reasons that are not yet understood, the affected members of the Pakistani families had no symptoms of a disordered sympathetic nervous system, such as irregular heart rate, and seemed entirely normal apart from the occasional self-inflicted damage caused by their inability to feel pain.
Though we don't know the precise genetic reason for this family's insulation from the normal debilitations which follow from their condition, theoretically I think it is very likely to be a modifier gene which lurks in the clan's genetic background and is not generally present in others who exhibit this mutation. If individual X carries mutation Y which should result in a decrement in fitness Z, but does not, I think an a priori plausible hypothesis is that there are other loci which mask the deleterious affects. Many single locus Mendelian diseases with moderate or low penetrance may simply be polygenic in nature and exhibit variation which is cryptic because those without the ailment aren't tested for the mutation.