It's not clear in this case what the difference is between gene, gene expression, and gene encoding. What does gene mean if it does not imply either encoding or expression? If this is a disagreement about the semantics of encoding, surely it'd be easiest to express this in terms of encoding. If this is about gene expression, surely it'd be easiest to express this in terms of expression. "gene" largely has no meaning outside of encoding/expression.

It seems clear to me the article is indicating a distinction of encoding. I'm not sure where the ambiguity is.

Second: what are genes? A "gene" is an abstraction.

Let's first define a "genome". A genome is the complete dump of raw data of "what the DNA base-pair sequence string says" — before any coding or expression occurs. Whenever we sequence DNA, it's [parts of] this raw base-pair string that we get back — not the codons, not the expressed RNA sequences. And this is why we care — sequencing is the tool we have, so it's the lens by which we look at DNA. And that lens shows us the raw data.

When we talk about "genes" (or "SNPs" / other bioinformatics terms), we're basically talking about the ways in which that raw data we can extract through sequencing, relates to observed phenotypic changes. A "gene" is a particular part of a "genome" (DNA base-pair sequence) that can be uniquely identified as being the cause of interesting phenotypic consequences when changes are made to it. (Which in turn is how we figure out what genes are "responsible for" doing what.)

Note how these concepts, "genome" and "gene", both completely ignore epigenetics / gene expression. That's because these terms were invented before epigenetics was invented, and these terms are part of a model that uses a lens (DNA sequencing) that itself ignores epigenetics. Gene sequencing shows you the world of DNA as if epigenetics didn't exist.

Now let's talk about DNA.

If you think that epigenetics is about changing how DNA encodes information, then you might be thinking of "DNA" via the lie-to-children model, of it just being a long sequence of nucleotides.

But consider: why do chromosomes look the way they do — little X shapes? A long string of base pairs, on its own, has no reason to assemble into that macroscopic shape.

This is because "DNA" — which is really a shorthand for "the DNA complex" (i.e. a complex of multiple weakly-bound molecules that together form the chromatin of a chromosome) — is not just nucleotide base-pairs. The nucleotide base-pairs form one molecule (the deoxyribonucleic acid string itself); but then you've got other stuff. You've got histones — little tape-spools the DNA string is wrapped onto. You've got methyl groups — little markers hanging off particular places on the DNA string. You've got other stuff I don't personally understand/know about.

When lay-people talk about "DNA", they're really referring to the whole complex of molecules that makes up each one of your chromosomes.

"Genes" are just the bioinformatic data representation of the deoxyribonucleic-acid-base-pair-string part of that complex of molecules.

But gene coding, and gene expression, are both a result of what the other molecules in that complex are doing.

(Think about it: if gene coding + expression were encoded in-band by the DNA itself, then that information would be appear in DNA sequencing — and so we'd be unable to differentiate that information from changes to the DNA itself — and so we wouldn't even have a concept of "epigenetics", because it would all just look like "genetics"!)

Here's a picture: https://www.genome.gov/sites/default/files/media/images/2022...

Histones — those little tape spools — attract or repel each-other due to chemical modifications to the histones themselves. Two histones that "snap together", prevent the region of DNA "tape" between them from being physically accessed by the RNA polymerase enzymes that "read the tape" to produce RNA.

Methylation markers hang off of the initial "landing sites" for RNA polymerase enzymes (CpG dinucleotides — think "floppy-disk sector header"), and repel them.

When you sequence DNA, you ignore these transcription-silencing signals. You can picture DNA sequencing as "unrolling" the deoxyribonucleic acid off of its histone carriers, rendering them irrelevant; and then using molecular tools to read the sequence — tools that, unlike RNA polymerase, are not repelled by methyl-groups.

If you use more-modern techniques to capture this additional information about where these silenced regions are, and by what mechanism they've been silenced, then you get what we call an "epigenome" — which isn't a post-translated version of DNA, but rather sort of a "metadata track" that runs parallel to the DNA "data track."

(And you can, in theory, combine the two to calculate an "expressed genome." I don't think we've ever done that yet — partly because "whole-epigenome sequencing" doesn't yet exist in the way that "whole-genome sequencing" does; and partly because epigenetic metadata is probabilistic — with some modifications decreasing the probability of a region coding for something, rather than turning it off altogether — and so current approaches, that derive the epigenome "by reaction", would observe something like "weak bits" on a floppy disk — regions that read differently each time, requiring many passes to calculate a "flux strength" for each region and to find each silenced region's true borders.)

I am a biochemist. The term is absolutely used all the time, in the CS sense, in biology.

A gene is a unit of heredity.

The genome is the sum total of all genetic information in an organism.

Both are extremely loose terms. Individual sub-fields of biology/biochemistry may use more specific variations in specific contexts, but the definitions above cover all of those meanings.

(I.e. per my sibling post, there were jargon terms with these names that had one particular definition under an early, black-box mental model of heritability that is no longer used/favored; and there are new replacement jargon terms with the same names, that have a different definition under the more-modern mental model of germline heritability.)

After all, there are entirely non-genetic mechanisms of heredity. You inherit some of your mother's immune histochemistry through developmental exposure, and more through consuming breastmilk. Yet there is nothing that "codes" for this histochemistry; it's just a stateful process (mother's immune system) interacting with another stateful process (foetal immune system) — working almost more like a "colonization" of commensal bacteria than like gene transfer.

In terms of "heritable traits" — "nature vs nurture" — such effects appear squarely on the "nature" side. But they're not germline heritable. Implant the fertilized egg into a surrogate mother, and you get different histochemistry. Have a wetnurse feed the baby, and you get different histochemistry.

If these things are "genetics" — if the information carried by the state of immune cells in response to observing natal antigens is "genes" — then the term "genetics" has no useful meaning / is useless to talk about the thing we want to talk about when we talk about genetics. Where "what we talk about when we talk about genetics", is how an organism's breeding/germline ancestry, predicts phenotypic outcomes.

And if you agree that that is the thing we want to talk about — then epigenetic information isn't "genes" (under the germline-heritability model) either. Your epigenetics is hereditary, but only in the same sense that your immunohistochemistry is hereditary — being passed via exposure in a natal environment, rather than being "written into" the fertilized egg.

(Also — at least AFAIK — we don't even consider our nuclear mitochondrial DNA to be a part of our genomes. Even though it is passed through the germline, and is intensely important in many congenital metabolic dysfunctions. Because our mitochondria are not technically "us" — we might call them organelles, but they're still more like commensal prokaryotes in how cells treat them: dividing on a separate timeline from our cells; not assortatively assigned to daughter cells by spindle fibers during mitosis; not observed + checkpointed to ensure sufficient numbers exist for daughter cells before telophase; etc.)

Nit: chromosomes mostly don't look like little X-shapes except during cell division, which was historically a useful time to image them (I think because the dye couldn't get through the cell-nucleus wall but during cell division it has dissolved?). But then you are seeing _two_ chromosomes that happen to be bound at a centromere before that centromere gets torn into two by the cytoskeleton dividing the cell in half.

See, this is a non-starter. Sure it's coherent, but you've simply failed on step one to identify how humans communicate. Regardless of if you're referring to "coding" or "encoding" (I'm not sure of the semantic difference between these two terms, frankly, nor an understanding why this distinction is worth presenting without remediation) a gene is still a heritable trait, not a specific span of a specific encoding of a specific molecule. A gene is, colloquially, a heritable trait that implies no specifics about how it's coded/encoded into whatever substrate you choose. You're referring to a far more specific term most people won't recognize as meaningful, and you can either explicitly describe what you're discussing or accept that people will misunderstand you.

I see you're trying to advocate for specific terminology to agree on, but it's far more effective to meet people where they already are rather than trying to push jargon onto others.

If I'm wrong, that people specifically refer to subsets of a DNA encoding by referring to "gene", please educate me.

It is not quite right to say: “a gene is … a heritable trait.” This definition equates gene with trait. Yes that is how Mendel thought about his results since he could only see and quantify traits. But there are still a depressingly large number of protein-coding genes without linkage to traits.

> It is not quite right to say: “a gene is … a heritable trait.”

This is a reality of communication you need to deal with: that's absolutely how people will interpret the term "gene".

> Yes that is how Mendel thought about his results since he could only see and quantify traits. But there are still a depressingly large number of protein-coding genes without linkage to traits.

From my perspective, from my exceedingly humble opinion, "heritable trait linked by reproduction and not culture" is absolutely how the public at large understands the term "gene". The fact that scientists cannot successfully link protein-coding to linguistically-bound traits is an issue that scientists will have to work around when communicating about the specifics of heritability, genetic determinism, and discretization of individuals. "Gene" is simply not a term you can reliably link to DNA in the popular consciousness and is likely not worth the effort or money to redefine.

No. There are two distinct ways in which humans communicate.

• Humans create and use words, whose usage stretches and spreads and grows and can only be defined descriptively.

• Humans (doing professional/technical/scientific work) also define-into-existence jargon terms — the meaning of which is set in stone at the moment of creation.

Jargon terms are verbal handles we use to grasp parts of particular, well-defined mental models — models that are taught and learned as coherent systems of such terms; models that exist to enable their users to think rigorously, and collaborate without corruption of meaning/loss of rigor, to "get things done" in those professional/technical/academic contexts.

Jargon terms, therefore, can always be said — definitively, objectively(!) — to have been employed correctly or incorrectly in any given text, per the mental model in which the jargon term is defined.

Sometimes, professionals/technical people/scientists are dumb and silly, and attempt to invent a jargon term made of words, where those words, put together in that order, have such an obvious, novel, and useful lay usage, that it is inevitable that anyone who hears the term will already think they know what it means, before ever being formally taught what it means.

"Begs the question" is an example of this — there is something so uniquely useful about the lay meaning of that phrase (essentially "a situation that demands that an unstated question be asked"), that whoever named (or more likely, translated from Latin into English) that logical fallacy, made a mistake by choosing to use those words to name it.

But that situation is rare.

Most of the time, professionals/technical people/scientists make up a word all on their own, to be jargon, entirely for their own use. (Often they use Latin for this, as an intentional way to dodge the aforementioned problem of the word "seeming like it has an intuitive meaning" in any living language. But they don't have to. They can also just make up a nonsense word, perhaps one inspired by a Latin prefix or something — a word like "gene.")

And then, through a pipeline of scientific discoveries, to science journalism, to pop-science, to regular people just talking about how their aunt has asthma due to "that bad gene she got from her father" — somewhere along the line, someone screws up in how they're using the jargon term; and that screwed-up understanding proliferates.

But jargon terms — at least the ones that are part of mental models anyone still employs any more — must still be used to do work. And they cannot function to do that work, if people can't use them with the implicit understanding that they are communicating an exact, precise, well-defined idea. Such words are not merely words; they are professional tools. The professionals that use these tools, cannot allow them to be corrupted, to have their meanings diffuse.

And so, at least the professionals themselves, will always be taught that any jargon term they learn has one specific meaning — that it acts as a handle for a particular concept in a particular mental model.

And those professionals will insist on this strictness when working with any technical writer or editor.

And this strictness diffuses from there, into any editorial context where editors consider themselves to be professionals, working for a publication that expects professionalism — whether that publication is technical or non-technical.

In professional editing, is always expected that if any term (word or phrase) is used which is known to be a (unique, non-colliding) jargon term rooted in some technical/academic mental model, then that term should only be permitted if it is being used with its original jargon meaning. Any other handling of such a term is considered a usage error, effectively a typo.

(For non-unique/colliding jargon terms where lay usage exceeds jargon usage, the lay-usage is given represented by plain printing, while the jargon usage is represented by italic printing. This only ever comes up with phrases — "begging the question" again — because even the most ivory-tower academics know that overloading the meanings of single extant lay words is a stupid way to invent jargon.)

In other words, it's about "those who know" being a role model for correct usage by "those who don't know."

(Why bother? Because then people are more likely to come into the profession with an understanding of the concept that's at least compatible with teaching them the model; rather than having an understanding of the model so debased that they need to actively unlearn it before they can learn the correct model.)

Who is the grand arbiter of whether or not use this use is "misuses" or, perahs "correct", whatever that means—the professionals, or the lay people?

The less mutable part. The strongly heritable stuff. Genes are part of your body, as is your brain, that doesn’t mean your every thought is equivalent to DNA.

(this does not mean that this DNA is directly heritable, since it does not necessarily end up in the gametes, but it is still interesting that, yes, in fact, it involves DNA change)

This also seems to be a semantic distinction largely irrelevant to understanding the article as presented.

I think neural nets show that, realistically, evolution requires backpropagation (epigenetics).

A good way to think of this is cell differentiation. Every cell in your body, except eggs and sperm, contains a copy of your entire genome but what differentiates a nerve cell from a liver cell from a skin cell is gene expression. All it takes is some changes to the epigenome of a stem cell during mitosis for one the resulting cells to become a new, differentiated cell.

DNA methyl-groups, acetylated histones, etc — these are the config file generated by running `menuconfig`. They determine:

1. which components of the kernel the "compiler toolchain" (RNA polymerase) will actually "compile" (transcribe) into kernel-module objects (RNA sequences, proteins) — in turn determining which "features" (developmental or metabolic processes) each of the components (cell/tissue types) of the resulting system (organism) will have; and

2. which #IFDEFs (coding regions) within those modules will actually be macro-enabled (expressed) during compile-time (transcription) — in turn influencing / varying / tuning the strategies and logic (frequency of production / likelihood of expression, final shape / folding / binding affinities of proteins) used by any given component (cell/tissue type) to perform a given "feature" (developmental or metabolic process).

(And just a tweak to steer this intuition pump — your body doesn't have one kernel, compiled once at conception; rather, your body is more like a distributed system composed of machines running Gentoo — every cell has its own kernel, and each cell is "tweaking" and "recompiling" its kernel regularly. When cells undergo mitosis, both daughter cells inherit these tweaks — that's like setting up a new one of these machines by mirroring the hard drive from an existing machine, carrying over not just the kernel but also the at-snapshot version of the `menuconfig` configuration file.)

oh my god would you shut up already

The Information Theory of Aging

The Linux kernel is just a much closer analogy than your average piece of software would be, because the Linux kernel actually has this distinct step that generates a standalone build config file that later build steps read.

Most regular software build infrastructure (autoconf, cmake, etc) just directly "burns in" any passed-in build config as generated code, without any intermediary file storing the config itself — which is a very different and not-very-analogous process. (That'd be more like if our epigenome was an extra chromosome of DNA built up during development!)

My guess for why this keeps happening, is that it's a two-step process, fueled by a failure of communication:

1. The science writer themselves does understand epigenetics — but doesn't think it's important to the point the article is making for the reader to understand epigenetics. The writer wants to remove the requirement/assumption of "understanding epigenetics" from their writing, while still being technically correct in everything they say. So they choose to gloss an epigenetic change as "causing changes to the DNA." (Which it certainly does! Either chemically — to the DNA molecules themselves, through methylation; or structurally/topologically — through modifications to the histones around which the DNA is wrapped.)

2. The science writer's not-so-scientific editor comes along, doing a stylistic editing pass; sees the word "DNA"; and says "hey, that's jargon, and we're aiming for accessibility here — we need to replace this." And they (incorrectly) decide that a valid 1:1 replacement for "DNA" is "genes" or "genome."

This invalidating change could be caught... if the publication had a formal workflow step / requirement for the editor to perform a back-check with the original writer after copyediting + stylistic editing, to ensure that validity has not been compromised. I believe that big-name science journals and science magazines do tend to have these back-check steps. But smaller publications — like the PR departments of universities — don't.

Unfortunately, they are just choosing to misrepresent the research sometimes.

That's not a valid defense, otherwise they could write with a straight face that sun exposure causes genetic changes that make you darker, and how your descendants will inherit those changes too. It is dishonest writing, bordering on disinformation. Said article would attract clicks and attention like a cursed magnet, though.

Human systems are imperfect and messy. Understanding how they break down helps us make them better.

It’s inexcusable that the quality of science communication is so low.

“…may encourage policymakers and humanitarian agencies to provide targeted resources to vulnerable populations…”

Hard to argue with that, but do we now use a handful of epigenetic markers generated using a cheek swab sample to decide who gets “healthcare access, special lodging, sanitation, and nutrition”? No, never.

One could get a lot further a lot faster doing well controlled experiments in animal models. But even here the sociological context intrudes. Mike Meaney’s original work in the early 2000s is still open to questions.

I don't think the activism/sociological angle here was supposed to be interpreted as being about targeted reconciliation toward existing victims of violence / their descendants. Even if those people are potentially "living life on hard mode" as a result of previous violence. (After all, finding and helping those people would require complex infrastructure and lots of money.)

Rather, I think the thrust of the implicit activist argument would be closer to:

1. Experiencing violence turns out to be even worse for you than we thought — and in non-recoverable ways! (You can't therapy your way out of epigenetic changes.)

2. And so more should be done to prevent/ameliorate violence and its impact on people, before it can cause durable epigenetic changes;

3. ...and some "low-hanging fruit" in reducing this impact for minimal cost, would be to supply more resources (e.g. shelters; crisis lines; social workers; restraining orders that actually work) specifically to groups who are demographically most likely to become victims of violence.

4. And maybe we can even use this correlational data to measure interventions like these, sampling the epigenomic changes in these vulnerable populations over time (repeated cross-sectional study), to know if marginal increases in certain types of resources actually have measurable impacts on "cushioning" these populations from violence, enough to prevent these epigenetic changes.

Use this (faulty) explanation to make a statement about the impact of violence -> direct resources to one demographic or another based on that (faulty) explanation of why it's important.

Your fourth point is entirely valid, if we were performing these measurements & experiments at a large enough scale to have a valid baseline in the first place (which GP was arguing is not the case).

Incorrect. The parent comment was talking about using epigenetic measurements of specific people to decide if they get resources. The child comment was not talking about doing that, instead saying that this research can be used to argue for more resources in general.

The only mention of demographics in the child comment was about groups likely to experience violence. Those groups would have no markers. It would be purely about how we can most effectively reduce violence.

The only measurements suggested in the child comment would be for statistical purposes.

If you want to direct resources to groups "likely to experience violence" based on other statistics anyway, then you don't need this type of research to do it. And if you're going to use this type of research in support of that, then it is just as important that it's accurate as it would be if you were using it at an individual level.

The parent comment called out "puny sample sizes." That would be relevant to the "statistical purposes" suggested by the child comment.

It's bad in a very different way, with a different level of impact. And there are ethical issues with that kind of use even if the research is right which don't exist with the broad statements they were suggesting.

> If you want to direct resources to groups "likely to experience violence" based on other statistics anyway, then you don't need this type of research to do it.

You do if you're trying to convince people to allocate the money, because longer lasting damage makes it more important.

> And if you're going to use this type of research in support of that, then it is just as important that it's accurate as it would be if you were using it at an individual level.

It's not. The consequences of screwing up individual diagnoses are much worse than the consequences of a funding campaign having a false premise.

Isn't that exactly what the study purports to show? Why would psychological trauma (i.e. negative nurturing) effect epigenetics but not psychological healing (i.e. positive nurturing)? What we consider "trauma" by modern standards would have been experienced by a huge fraction of every prior generation, and over the course of generations the entire population countless times over via epigenetic inheritance. Either it works both ways, even if unevenly, or it's BS (the framing if not the science). Ideological motivations aside, it just happens to be much easier to identify and track discrete incidences of severe trauma.

Yes, you can cushion people from the effect of trauma as it's happening, by providing resources at or soon after the point of trauma.

But your epigenome is a record (almost exclusively) of stuff that was observed during your development — it's a log, not a state database.

A particular epigenome-affecting gene process would look like: "when I was -6 months old, a process kicked off that began expressing a gene. This gene caused the expression of a protein that acted to detect whether there was enough magnesium in my blood or not. It detected that there was enough — and so the protein recorded [by not methylating anything] that I shouldn't silence any of the copies of the magnesium-transporter in my intestinal-lining tissue cells. Then, the buildup of the protein caused the expression of another gene, which methylated [silenced] the first gene, permanently deactivating this magnesium detection step from happening again."

Which means that trauma during early development — if it is not cushioned at the time, such that its impact is felt fully* — will lead to some epigenome-affecting process that looks for "impact of trauma" observing said trauma as happening at the time it activates, and so recording it. That record, within the cell-line descending from the cell that made that observation, is now permanent; that tissue will now do whatever it does with that knowledge, from then on, for the rest of the organism's life.

Treating the trauma later, will help the organism psychologically, but will not change those markers (and whatever biological effects they have), because they are markers of "what was at a certain point", not "what is."

That's a leap. After all, emotionally distressing experiences presumably caused the epigenetic changes to begin with, so why couldn't other mental experiences alleviate them? Not that we know how to do this at this point, I'm just stating that your claim is not a fact.

Crucially, because these "setup programs" silence themselves after running, these "setup programs" each only have one opportunity to run, at some genetically-defined point in an organism's lifecycle. The information is only collected once. Whatever this "setup program" observed at that point, that's what the epigenome records. There is no second chance to set it differently.

I don't know much about this violence finding, but one famous finding in epigenetics is that mammals whose mothers were often food-deprived during gestation, end up with a marker that correlates with the phenotype of eating more / being more food-motivated / tending more toward gathering more food than is necessary for their immediate satiation.

At least in the case of this food-deprivation finding, there is nothing you can do to/with the animal afterward, that reverses the impact of this epigenetic change. The animal's body doesn't go back and re-evaluate whether or not it "should" be more food-motivated, because the fact it's basing its food-motivation calculations off isn't about what's true now — it's about what was true then. The marker doesn't mean "the environment is currently food-poor", it means "the environment was food-poor when I was -6mo old."

(In fact, these early-natal-development epigenetic markers aren't even necessarily what you might call runtime flags — they're more like an #IFDEF statement in the DNA "codebase", their values first calculated during the long, continuous "build process" of development, and then macro-expanded later in that same build process to hard-code other things — e.g. how certain proteins/tissues and their chemical receptors get built. By the time a baby is born, that marker might actually have no further relevance, having done all its work and made its downstream impacts during gestation. If we had "epigenetic CRISPR" and turned that marker off in an adult person, doing so would only impact whatever that gene does after development — which might be "nothing.")

I just want to add the obvious addendum again: that we know of. We've tried a few obvious things and that didn't change the behaviour, but that doesn't mean much in something this complex.

The argument that "brains are rewired" or "generations of genetics are harmed" is popular in 'trauma informed' arguments for compensation or government help.

*edit

Anyone downvoting me should go read this study. The number of individuals exposed to the initial dose is in the single digits. There is no attempt to account for confounding variables. No mechanism is known that could account for the proposed effect. The preceding animal study they referenced has been HEAVILY criticized for methodological issues.

If you're speaking from an american perspective, this hand-wringing is unnecessary: we leave our own citizens exposed on the street. If we were capable of making decisions about who deserves xyz in the first place we wouldn't be living in such an obvious shithole to begin with. Maybe we should set our standards much lower if we aim for research to actually benefit humans.

I read about DNA methylation (well, epigenetics in general) in one book for behavioral epigenetics.

You seem to be disappointed in the headline because you're expecting narrower definitions.

This is an extremely pedantic distinction, especially since the article as a whole is pretty clear.

In nearly all cases, what matters is whether we're talking about information that is encoded and passed down. That's the case here, even if the encoding and persistance characteristics are different.

Crucial difference, one is a mutation, the other an environmental adaptation.

It's in the title and it's clear throughout the text of the article.

So that's a totally moot point.

  Mulligan CJ, Clukay CJ, Matarazzo A, Hadfield K, Nevell L, Dajani R, Panter-Brick C. Novel GxE effects and resilience: A case:control longitudinal study of psychosocial stress with war-affected youth. PLoS One. 2022 Apr 4;17(4):e0266509. doi: 10.1371/journal.pone.0266509. PMID: 35377919; PMCID: PMC8979449.
  
  Clukay CJ, Dajani R, Hadfield K, Quinlan J, Panter-Brick C, Mulligan CJ. Association of MAOA genetic variants and resilience with psychosocial stress: A longitudinal study of Syrian refugees. PLoS One. 2019 Jul 17;14(7):e0219385. doi: 10.1371/journal.pone.0219385. PMID: 31314763; PMCID: PMC6636744.
  
  Panter-Brick C, Eggerman M, Ager A, Hadfield K, Dajani R. Measuring the psychosocial, biological, and cognitive signatures of profound stress in humanitarian settings: impacts, challenges, and strategies in the field. Confl Health. 2020 Jun 23;14:40. doi: 10.1186/s13031-020-00286-w. PMID: 32582366; PMCID: PMC7310257.
  
  FAAH, SLC6A4, and BDNF variants are not associated with psychosocial stress and mental health outcomes in a population of Syrian refugee youth
  Christopher J. Clukay, Anthony Matarazzo, Rana Dajani, Kristin Hadfield, Catherine Panter-Brick, Connie J. Mulligan
  bioRxiv 685636; doi: https://doi.org/10.1101/685636

AFAIU, the later mostly encodes the structure of proteins, which naively are the factory machines of or cells (enzymes, that is) and some of its building blocks. The gene expression tells the factory what to produce (a tail or an ear, say), when to produce it, and how much. More relevant to the topic of this article, the gene expression would determine which structures and pathways of the brains (or adrenal glands, or any other organs) to suppress, and which to reinforce and build up in a new generation of humans.

The building blocks are important, but once they are good enough, it matters more what you choose to build with them.

As in after x generations the genes "go back to normal"