Early humans domesticated themselves, new genetic evidence suggests


Really excellent study. Thank you for posting @Patrick.

1 Like

Could a biologist (@swamidass or others) expound on this just a bit:

When the researchers looked at those hundreds of BAZ1B -sensitive genes in modern humans, two Neanderthals, and one Denisovan, they found that in the modern humans, those genes had accumulated loads of regulatory mutations of their own. This suggests natural selection was shaping them.

Why does this suggest natural selection and what is the significance of that?


Great question for @Joe_Felsenstein, @John_Harshman, @art, and @NLENTS.


An unusually large of number of changes in functional elements in the same pathway suggest that something is unusual about that pathway. Either there is sustained positive selection acting to change the biological behavior of the pathway, or the pathway has become less important and is under relaxed purifying selection. In this case, the pathway is a developmental pathway involved a number of notable features that distinguish modern humans, including social traits and facial morphology. So it’s a good bet that evolution of this pathway was responsible for some of our unique features, and that they are the result of natural selection. What is not obvious is what specific trait or traits were being selected for.


Is this evidence that Sapiens are more “human” than Neanderthals and Denisovans?

Let me give this a try.
We can infer recent natural selection in a gene (or regulatory element) by looking for recent mutational changes that cluster in a nonrandom way. Consider a stretch of DNA 1k base pairs long. If that DNA was serving no important purpose in the organism, it would not be subject to natural selection and would therefore accumulate mutations at a more less steady pace and randomly distributed throughout the entire stretch. (The fixing of mutations is actually the subject of genetic drift, which is important to understand this also, but I’m trying to keep this simple and focused.) However, if you see mutations clustering in specific places, that suggests natural selection. For example, some of the micro-RNA genes that I am currently studying show this. They are very small, like only 300bp. We are comparing the DNA of all the apes at that specific region. (Apes = humans, neanderthals, chimps, bonobos, gorillas, orangutans, and gibbons - yes we have the full genomes of all of those species!) What we find is that if you look at DNA upstream and downstream of that gene, you get the predictable degrees of similarity. Humans and chimps are 98.5% similar. Humans and gorillas, 95%, and so on. However, if you look just within the 300bp micro-RNA gene, the similarity is way less. Humans and chimps are 91% similar. And if you look at the precise region of the gene that is most crucial, the stem-loop, it is down to about 80% similarity. What this means is that this gene has been rapidly evolving in the human lineage. (Or the chimp lineage, but in this case, it’s the human gene that’s the outlier. The chimp version is the expected similarity to the gorilla version and so on.) The analysis is a bit more complicated than this, but I think that summarizes the logic of it well.

Now, that’s a simpler case because it’s an RNA gene. Genes for protein, like the article above is talking about, are more complicated, but also allow greater granularity. Within a protein-coding gene you have introns and exons, noncoding and coding regions, which are subject to different levels of scrutiny by natural selection. Simply comparing mutational rates between those two can sometimes tell you something. Focusing further, you can look at synonymous vs. non-synonymous mutations (whether they change the protein sequence or not) and you can can even focus on specific amino acid positions in the protein because some are more identity-crucial than others.

This work shows how sequencing the genomes of other animals doesn’t only tell us about that animal, but can sometimes help us understand the history of our own genome. I study the emergence and evolution of “young genes” in humans and I couldn’t do my work without access to the full genomes of the other apes. (You can also sometimes detect recent natural selection by looking at variation among current human populations, btw.)

EDIT: Oh, but looking closer at the paper above, I see that they were focusing on mutations in regulatory regions. Same principle, but they found the recent evolution not in the protein-coding regions, but the regions surrounding them, which influence when, where, and how much the genes are expressed. While evolution of proteins is more “exciting,” most evolutionary change is this kind of subtle tweaking of regulatory information.

Another edit: I also thought of something else that can be important. Going back to our example of gene that has undergone recent rapid evolution, that’s actually the opposite of what we see in most genes. Between humans and chimps, most genes have NOT been evolving much, and so what we see is even greater similarity than we would expect by chance. If a stretch of pointless DNA is 98.5% similar, a gene that is very important and has not been evolving much, we see even greater similarity, approaching 100%. Within a given gene, you often see that the introns are conserved between humans and chimps at around 98.5% and the exons are more like 99.5%. The older the gene, the more conserved it will be (as a general rule). Take the beta-globin gene. Nature has been tinkering with that gene for so long that it’s pretty much as optimal as it can be without managing some wholescale redesign which, as we know, is not how evolution works (at least not very often). For this reason, almost any possible change reduces function and so you see nearly 100% conservation in the coding regions of the human and chimp beta-globin gene. When you step just outside those coding regions, however, you’re back to the expected 98.5% that random genetic drifts gives us. (Don’t take these numbers too strictly. They’re averages. Like all random systems, you see variation because its… um, random. lol)

How did I do?


You did great! I’m trying to build up my genetic “intuition” about things.

Initially, I was wondering how something could be changing faster than the mutation rate as I am used to discussions about natural selection conserving sequences by “weeding out” mutations.

But then I remembered it was an equilibrium, so it would make sense if the “good” changes were “picked out” by natural selection within the flow changes created by mutation, then it would be changing faster than the general equilibrium rate.

It is interesting to me that both lower and higher mutation rates can indicate selection. The difference is how far down the path towards optimization that sequence is, and presumably also how important that sequence is to the actual organism?


Right. Selection can be stabilizing or directional, which presents as lower or higher mutational rates respectively. (There’s also disruptive selection and other layers of nuance, but we’re talking generally here.)

There are very simple selection simulators out there that help visualize this. I remember having my mind blown as an undergraduate when I saw that the tiniest degree of natural selection, like a 1% advantage, can result in allele fixation in just a few dozen generations. Selection is really powerful once you escape from thinking about life in single-digit numbers of generations.


Really great job making the complex accessible!! Thanks Nathan!!