Constructive Neutral Evolution

For some reason I didn’t notice this last part of your post earlier. There was a thread devoted to a recent review article discussing what is currently known about mechanisms of so-called de novo gene evolution here:

It looks like this work is still highly theoretical. Would you agree?

I wouldn’t agree at all, but “highly theoretical” is a phrase with nearly unlimited wiggle room.

Here are two recent publications on de novo gene birth that are not “theoretical,” if by that you mean based on simulation or discussing theories.


It’s not that there aren’t aspects of the subject for which that isn’t true, but I’m weary that a two-word summary like that is bound to serve as a means for some to dismiss the entire subject as nothing but unsubstantiated speculation, so to those who would take it like that, I wouldn’t really agree with it. One has to say more than just that to do it any justice.

There are many genes for which we don’t know how they originated, and it’s hard to say which of the numerous known mechanisms by which novel protein coding genes can evolve, that have historically predominated.

But we certainly know enough already to be able to say with substantial confidence that it is a real phenomenon, that novel protein coding genes can originate by characterized evolutionary mechanisms, and that this has contributed many, many novel proteins over the history of life on Earth.


I agree, that phrase is probably too dismissive sounding.

Thanks, I’ll check these out.


That’s fair. I can see “highly theoretical” being a somewhat loaded phrase. Perhaps it would be more accurate to simply say there is still a fair amount of speculation going on.
To be clear, I’m not trying to dismiss anything, I care about the truth regardless of the implications. I just have not been particularly impressed with some of the sources provided. They seem far from certain and include a lot of vague qualifying language.


That’s how you can tell that you’re reading real science and not propaganda.

But if you are looking for something that is “certain” about the deep past, then you needn’t read any more.


Not looking for proof, just evidence from which we can draw reasonable conclusions (plausibility). Of course, what constitutes evidence is a personal value judgement (to some extent).

Regarding totally novel de-novo genes, I have not read anything that was particularly compelling. I am open minded though, and perhaps your source may prove more convincing.

I’m curious, what portion of the human genome would you estimate to be non-functional, or “junk”?

Any good material on dGRNs and how they might be modified to enable new body plans?

1 Like

What evidence is there that any de novo genes are totally novel?

That isn’t a technical term. “Function” is relative term. By some definitions nearly all of it is functional. By others, less than 2% is functional.

1 Like

Again, those are loaded wiggly terms and I have written about the various ways that someone might assign “function” to a section of the genome such as an intron. Until a person says what they mean by “function,” the question is hard to answer.

We know that vast swaths of the human genome are composed of repeats, and we know that millions of these repeats are composed of the remnants of mobile genetic elements. We also know that introns make up about 25% of the human genome, and we know that very little of that sequence space can be assigned a “function” other than as a speed bump. So, for me, I think that at most 50% of the human genome can be rationally assigned a “function,” and I would lean more toward an estimate of 75% or a bit higher as “non-functional.”

Yes there are whole subdisciplines that study this, assuming that dGRN = developmental GRN. That’s evo-devo and you should be able to discover good work with ease.

1 Like

Well what determines what you find compelling? It’s hard to grapple with an incredulous stare.

Unavoidably, inference of de novo gene evolution has to involve historical inference. Nobody was around to record what happened in the genomes of some population of organisms hundreds of thousands, or millions of years ago. If this is going to come down to not seeing it happen in real time (or that no other explanation should be possible to conceive of), you will never find any inferences about the past compelling. I assume this is not the situation we are in though.

So the alternative we have, since we are forced to do historical inference with the comparative data we have, is to compare the predictions of models to observations. And to invoke observed processes and phenomena to explain that data. Same as we would do in any other branch of science that involves historical inference, where scientists attempt to explain data by invoking observed phenomena to explain patterns we find.
In this sense I think it can be valuable to reflect on the fact that nothing is happening in evolutionary biology as a matter of explanatory principle that sets it apart from physics, astronomy, paleoclimatology, or geology.

The basic principles of historical inference are the same, so it’s difficult for me to understand why you’d find some putatively good case for de novo gene evolution (such as, say, the BSC4 gene in yeast) any less compelling than you would a historical inference about how plate tectonics yield mountain ranges.

Most models basically conform to this scenario: (Figure from here: Fact or fiction: updates on how protein-coding genes might emerge de novo from previously non-coding DNA)

It is then a matter of finding a putative taxonomically restricted gene, and determining by genome comparison if a similar sequence of events is reflected on a phylogeny of the species carrying the gene of interest. If scientists can do that (and they can, and have), then there really is good evidence for de novo protein coding gene evolution. I just don’t see what the real problem is here. What more could one hope for, and why is that necessary?


@moderators, the de novo off-topic should be carefully split because it is interesting in its own right.

Likewise, much of the conversation with @ETVB is quite helpful and I see him as asking legitimate and good faith questions, so it might be worth splitting into additional threads too.


It’s not just this. Imagine a totally neutral region of the genome. It would be very effective at the “function” of recording history uninfluenced by selection. “Function” is almost entirely in the eye of the beholder.


Has anyone looked at taxonomically restricted DNA sequences? That might be far more difficult, and far more interesting.


1 Like

And that’s why the conversations about “function” almost invariably involve religious apologists and not scientists.

If a long intron slows down gene expression (it does) and this is shown to be important functionally in a specific case and context, then do we say that the 603rd base pair of that intron has “function”? Questions about how introns affect gene expression are interesting. Questions about whether introns have “function” are not interesting and are usually red herrings.

Oh BTW I’m rather proud of my use of the Yugo as an illustration of all of this. Here’s the relevant excerpt, edited to replace ‘lie’ with ‘falsehood’:

The second aspect of this grand creationist falsehood is the claim that “junk DNA” is functional. And here’s where things get a little sticky. Look carefully at a typical creationist argument of this type. It goes like this:

  1. Evolutionists say (or said) that “junk DNA” has no function. (This is a falsehood, but we already covered that.)
  2. But here is a paper describing a non-coding DNA element that has a biological function.
  3. Therefore, “junk DNA” is functional.

The error here is pretty simple, but I think the argument typically exerts its influence first by virtue of the dishonest first premise, and then by an impressive-sounding (and usually perfectly accurate) discussion of a recent discovery in biology. The bogus conclusion is thus easier to smuggle in, especially if the audience isn’t thinking carefully or is otherwise overly credulous.

Here’s how to see the error (if it’s not already obvious):

  1. Insurance companies say that any 1989 Yugo is worthless and has no utility of any kind. They were crap when they were new, and they’re worse than that now.

  2. But let me tell you a story about a 1989 Yugo that is being used as a perfectly good mailbox (or church confessional, or shower).

  3. Therefore, 1989 Yugos are valuable and useful.

Think about it: that first statement can’t be right. Insurance companies probably don’t say it just like that, and certainly they don’t mean that an old Yugo can’t be put to some good use. If you showed them one that actually runs, and proved that you drive it to work once a week, they would happily admit that it has some value. But who cares about that anyway? It’s a red herring, and it’s wrong to boot. Just forget about that silly first claim for a second, and follow the rest of the argument.

It’s plain ludicrous. Of course some 1989 Yugos are valuable and useful, but that hardly means that 1989 Yugos are generally valuable at all. If you met someone who asserted that 1989 Yugos were “functional,” and who claimed that those who say otherwise are involved in a nefarious conspiracy, you would probably take careful note of the locations of the exits.


I’m writing an article describing our discovery of a human-specific gene and our hypothesis about how it evolved.


As we’ve already seen, as scientists fill in the gaps and find evidence for the emergence of de novo genes through genomic rearrangements/duplications followed by natural selection, ID proponents will simply retreat to their safe space of saying, “Okay, but you can’t show that the rearrangements were unguided.” Notice the subtle shift from saying it couldn’t happen, to it couldn’t happen without guidance, which would leave no evidence to be found anyway. Ultimately, these arguments always end up back in the same place… neither scientists nor ID proponents can show exactly what force causes a specific mutation (or duplication or anything else), and somehow that’s a point against evolution and for ID. Heads I win; tails you lose.


True, functions are relative, and it sounds like most mutations are essentially non-detectable/non-selectable. However if biological information mediums work similarly to the way languages do, information degrades with more scrambling. In theory you could get what appears to be new information, if there were pre-programmed “sub-routines” set up for adaptation to various circumstances. So how can we tell if a gene is de novo, or part of a design?

I do. No need to be rude.

I believe I cited literature that explains how de novo genes are detected. By comparing the sequences from different species on a rooted phylogeny (thus enabling you to infer a direction of change), and then detecting the gradual mutational transformation of non-coding into coding sequence. And then by biochemical characterization of the function of the novel gene and it’s effect on reproductive fitness.

Again, it’s hard to imagine what else you could possibly hope to be provided of evidence for a historical evolutionary trajectory. There are numerous good examples where such work has been done and the mechanism of historical de novo gene evolution has been elucidated.