Biology Stack Exchange is a question and answer site for biology researchers, academics, and students. It only takes a minute to sign up.
Sign up to join this community
Consider the following statement
All polygenic traits involve epistasis.
I think its true because polygenic traits involve several genes interacting together to result in a certain phenotype and if one of these genes is altered, the resulting phenotype is different.
A polygenic trait is a phenotypic trait for which the variance in the population is explained by two or more loci. Epistasis is a case there is a statistical interaction between several loci on the value of a phenotypic trait.
The several loci underlying a polygenic trait might be purely additive (no interaction) and therefore a polygenic trait does not necessarily involve epistasis.
This is a sort of weird question, because it is so vague, but it is still important because quantitative genetics seems incapable of giving a coherent answer even for people with PhDs.
The idea of "additivity", for instance, is a statistical rather than biological concept. It was first applied to quantitative genetics by Fisher in his 1918 paper, and it comes down to a neat trick developed for predicting the phenotypic correlations between relatives (what we would now call "heritability"). Fisher showed that if you assume that genes are roughly additive/independent in their contributions to traits, that makes the math much simpler for calculating heritabilities. He goes on to argue that the approximation works ok as long as you let the number of genes tend to infinity.
More recent research, notably by Peter Visscher and co, suggests that the additive approximation is in fact pretty good for all measured human traits for which there is data. So, this would argue against a big role of epistasis.
The problem with all this is that there is no reference to biology whatsoever. Anything that happens in a cell is governed by interactions, e.g. epistasis. When two proteins work in the same pathway, their genes have an epistatic relationship. So, anything that happens in a cell is entirely dependent on epistasis. In fact, in highly controlled experiments where you can directly measure the epistatic component by measuring all possible genotypes, you generally find that epistasis is a really important component of variation, because additivity is just a best guess at the sum of all epistasis effects in a sample population.
Visscher and co have argued back that while all this epistasis stuff could be interesting, Occam's razor means that additivity is probably good enough for the purposes of quantitative genetics. Yet more recent work has suggested that you in fact improve predictions a bit if you consider pairwise epistasis, so it's very much an active field of research.
In summary: this is a question that sort of depends on where you fall on the experimentalist-to-statistician continuum, with the experimentalists saying it's all epistasis because look at their data showing epistasis, and the statisticians saying that additivity is probably ok on its own because look at the quality of the predictions. Epistasis is important if you care about things that happen inside of cells and organisms but not so much if you just care about easy phenotypic predictions.
