In genomics there are lots of non-variable base positions, because if a mutation were to occur there it would be deleterious. But of course there are many polymorphic sites that are deleterious too, and a large number that are potentially selectively neutral or functionally neutral.

An old idea for which GWAS data is now providing new evidence is that complex traits follow the infinitesimal model - where an infinite number sites each with an infinitesimally small effect exist. Of course this is not to be taken literally but the point is, is it the case that a very large number of methylation sites influence complex traits, each with a small effect?