I don't think it should be surprising that most genetic effects combine roughly linearly. If the effects were highly nonlinear then genetic selection wouldn't work well (just ask anyone who's tried to maximize highly nonlinear functions of many variables).
So I think a lot of people, me included, have an intuition that building a body from DNA must be highly non-linear. But I’ve heard the answer from statistical geneticists that the process of recombination rules out non-linearities or make them less likely. The intuition is that if you need two separate variants to get a selective advantage, then it is half as likely as usual that your children inherit both variants together. So over time any selective advantage is likely to be lost.
That would change if complementary variants were more likely to be inherited together, for example if they were close together on the genome. But then presumably that would also weaken the logic in your article, that non-identical twins are less than 50% likely to have inherited both variants.
I think you’re talking about this point when you mention “the epistatic variance is converted into new additive variance, which in turn increases the response to selection and so on for many generations”. But I didn’t quite understand this part.
See Old World monkeys where in most species only female heterozygotes have trichromatic vision. Unlike us, they didn't get duplication of longer-wavelength color receptor and therefore stuck in situation where strong selection pressure can't make all of them have trichromatic vision.
I don't think it should be surprising that most genetic effects combine roughly linearly. If the effects were highly nonlinear then genetic selection wouldn't work well (just ask anyone who's tried to maximize highly nonlinear functions of many variables).
Less of a problem if two genes are close to each other on a chromosome, since then they are more likely to be inherited together, correct?
Wonder if it's worth looking harder for epistasis specially for gene pairs that are close together on the same chromosome.
Been waiting for this, great discussion!
So I think a lot of people, me included, have an intuition that building a body from DNA must be highly non-linear. But I’ve heard the answer from statistical geneticists that the process of recombination rules out non-linearities or make them less likely. The intuition is that if you need two separate variants to get a selective advantage, then it is half as likely as usual that your children inherit both variants together. So over time any selective advantage is likely to be lost.
That would change if complementary variants were more likely to be inherited together, for example if they were close together on the genome. But then presumably that would also weaken the logic in your article, that non-identical twins are less than 50% likely to have inherited both variants.
I think you’re talking about this point when you mention “the epistatic variance is converted into new additive variance, which in turn increases the response to selection and so on for many generations”. But I didn’t quite understand this part.
See Old World monkeys where in most species only female heterozygotes have trichromatic vision. Unlike us, they didn't get duplication of longer-wavelength color receptor and therefore stuck in situation where strong selection pressure can't make all of them have trichromatic vision.