Ewan Birney

(it's a somewhat sad tale; the scientific community has confused itself about meaning of epigenetics and its consequences or 'paradigm breaking' aspects of it, meaning that the broader understanding is unsurprisingly a mess)

Epigenetics has come to mean one of three phenomena (1) the ability of cells in multicellular organisms to have 'state', in particular through cell division - a liver cell 'knows' it is a liver cell, even when taken out of a liver, cultured in the lab.

(2) Complex patterns of inheritance where parent of origin, haplodiploidy (Ants etc) and other molecular shennanigans makes inheritance in one (or sometimes two) generations complex.

Folded into this is the mammalian scenario where the eggs (Oocytes) develop early in the female fetus, meaning that every mammal has a direct molecular connection to DNA which was present in their grandmother

(3) Is complex pieces of inheritance that does not use DNA and goes over >2 generations. This trans-generational epigenetics is (relatively) common in plants, and rare in mammals (only one real case - Agouti mouse coat colour).

All three phenomena are (a) important and interesting and (b) often due to modifications of DNA with complex molecular roles of histones and RNA, and can now be (relatively) easily measured ("epigenomics").

Now: none of this is a "radical challenge" to genetics but rather very much (in my view) a part of the whole bit of understanding life, alongside genetics. The first, cellular state, is an absolutely key part of how we understand development.

The second is fascinating and complex, and gives rise to all sorts of interesting things - for example, there are particular sheep traits which only come through the mother. Male pattern baldness is a human trait with this.

The presence of eggs in the mammalian grandmother means that there is a very direct connection of exposure from grandmothers to their grandchildren (literally - a key bit of DNA is present in the grandmother).

For me, this is the likely transmission scheme for example in the Dutch Famine cohort, where grandchildren of the cohort have measureable differences from the quasi-matched controls in other years

(it need not be; other things including complicated confounders could be the explanation, and this cohort is super-interesting to study but this is a credible mechanism)

Trans-generational inheritance not via DNA happens really quite a bit in plants, where it seems to be used as a "tuneable" mutation rate to get the right amount of "spread betting" for environmentally sensitive traits, eg, flowering time

Actually some bacteria do similar spread betting strategies of deliberately randomising a trait (in this case, to evade the immune system) but with a DNA not DNA modification/RNA mechanism.

Very good people have looked hard for transgenerational epigenetic transmission in mammals. Beyond Agouti coat colour in mice, noone has found something >3 generations and lots of just 1 or 2 generational (interesting; but not the same!)

There is an argument in mammals the presence of learnt behaviour from parents in many mammals negates the need for a tuneable mutation scheme - an interesting discussion

The "missing heritability" of common genetics is not really linked here. Most of the missing heritability is power issues between the fearsome need of samples for identification of rare loci, vs the robust ways to measure their aggregate population effect (eg: twin studies)

Rather reassuring the power calculations predict this gap, and this gap is progressively closing with more samples as predicted. There *is* still a gap, in particular for some phenotypes, but lots of other pesky technical reasons to have a gap

Furthermore the easiest explanation for the gap is "Broad sense" to "Narrow sense" heritability, eg epistasis, and doing this in humans has the very real issue of parental behavioural transmission being confounded mainly with genetics

(in other species one can control for this a lot better - eg, in Cows, the sperm is sent from the prize bulls to the cows).

I really doubt trans-generational epigenetics is the major explanation of the missing heritability gap, but this "pushing back" on epigenetics is IN NO WAY saying its not important. It is *critical* for development, and probably other things.