Latest / Razib Khan's Unsupervised Learning / 14,000 years of natural selection
Transcript
- 0:00SPEAKER_00: This podcast is brought to you by the Albany Public Library Main Branch and the
- 0:04SPEAKER_00: generosity of listeners like you.
- 0:09SPEAKER_00: What is a podcast?
- 0:16SPEAKER_00: God, Daddy, these people talk as much as you do.
- 0:26SPEAKER_00: Razeeb Khan's Unsupervised Learning
- 0:29SPEAKER_00: Embryos have a 30% to 60% difference in the predisposition to severe diseases.
- 0:33SPEAKER_00: With ORCID, parents can identify their embryo at lowest genetic risk rather than just a random outcome.
- 0:38SPEAKER_00: Our genetic calculator can show the impact of embryo selection.
- 0:41SPEAKER_00: Discuss embryo screening and IVF with a genetics expert.
- 0:45SPEAKER_00: Welcome to another episode of Unsupervised Learning.
- 0:48SPEAKER_00: This time I am also doing a monologue like I did last week.
- 0:54SPEAKER_00: Yeah,
- 0:56SPEAKER_00: I was not expecting to do this this week,
- 0:58SPEAKER_00: but there's a preprint that came out that people have been talking about and people
- 1:03SPEAKER_00: have been waiting on for a long time.
- 1:07SPEAKER_00: I think the main reason that people are talking about it is kind of like one
- 1:12SPEAKER_00: particular finding,
- 1:13SPEAKER_00: actually,
- 1:13SPEAKER_00: and the preprint is huge,
- 1:16SPEAKER_00: but one particular finding is the polygenic score for educational attainment and
- 1:25SPEAKER_00: household income,
- 1:25SPEAKER_00: and I think there was one measure for intelligence,
- 1:30SPEAKER_00: all increased in the ancient DNA sample in this preprint,
- 1:34SPEAKER_00: and so it showed that...
- 1:37SPEAKER_00: If you naively projected it,
- 1:39SPEAKER_00: Western hunter-gatherers from 8,000 BCE had an average IQ that was like 88 as
- 1:47SPEAKER_00: opposed to 100,
- 1:48SPEAKER_00: which is a standardized European norm.
- 1:52SPEAKER_00: It showed that people got smarter during the Neolithic.
- 1:55SPEAKER_00: I'm front-loading the sexy, spicy, controversial aspect, but I'm going to talk about that later.
- 2:03SPEAKER_00: Much later in this podcast,
- 2:05SPEAKER_00: partly just because,
- 2:08SPEAKER_00: you know,
- 2:09SPEAKER_00: really people should read the methods and the supplements.
- 2:13SPEAKER_00: And honestly, I like to read the supplements pretty thoroughly and I did not.
- 2:20SPEAKER_00: read them thoroughly.
- 2:21SPEAKER_00: I'm just going to tell you, uh, on this reading, uh, for this paper, because there's a lot of them.
- 2:26SPEAKER_00: It's pretty abstruse.
- 2:28SPEAKER_00: And, uh, I wanted to get this podcast out because people have been asking me, uh, what I think about it.
- 2:34SPEAKER_00: Um,
- 2:34SPEAKER_00: and so,
- 2:35SPEAKER_00: you know,
- 2:35SPEAKER_00: I'm,
- 2:35SPEAKER_00: I'm getting it out,
- 2:36SPEAKER_00: but,
- 2:36SPEAKER_00: um,
- 2:37SPEAKER_00: I'm probably going to read the supplements at some point.
- 2:42SPEAKER_00: Uh, and,
- 2:43SPEAKER_00: you know, in depth to, for me to be happy about it.
- 2:46SPEAKER_00: Uh,
- 2:47SPEAKER_00: so the,
- 2:47SPEAKER_00: so the preprint is pervasive findings of directional selection,
- 2:52SPEAKER_00: realize the promise of ancient DNA to elucidate human adaptation.
- 2:56SPEAKER_00: Um, and, uh, yeah, the authors are, uh, Ali Akbari, Alison Barton, Steven Gazal, Zhang Li, uh,
- 3:09SPEAKER_00: So Akbari is the first author, and...
- 3:32SPEAKER_00: Reich is the last author, so it's out of David Reich's lab.
- 3:35SPEAKER_00: And I met Ali Akbari.
- 3:38SPEAKER_00: We had discussions.
- 3:39SPEAKER_00: America studied human genetics.
- 3:41SPEAKER_00: I know he's been working on this forever,
- 3:43SPEAKER_00: for a very,
- 3:44SPEAKER_00: very long time,
- 3:46SPEAKER_00: out of the Reich lab,
- 3:48SPEAKER_00: where mostly they studied population structure,
- 3:50SPEAKER_00: phylogenetics,
- 3:50SPEAKER_00: population history,
- 3:52SPEAKER_00: phylogenomics.
- 3:53SPEAKER_00: But they didn't really start doing stuff on selection until the last couple of years.
- 3:58SPEAKER_00: You know,
- 3:58SPEAKER_00: I think Vagish Narasimhan,
- 4:01SPEAKER_00: who is now at UT here at University of Texas Austin,
- 4:04SPEAKER_00: he did some of the work and there was a couple of others that came out.
- 4:07SPEAKER_00: But this is really, and I think Ian Matheson, who's at Penn now, also had the right lab, like Vagish.
- 4:14SPEAKER_00: You know,
- 4:15SPEAKER_00: there were people that were,
- 4:16SPEAKER_00: stuff has been coming out of dribs and drabs,
- 4:17SPEAKER_00: but this is kind of the big one.
- 4:19SPEAKER_00: And they attached it with a selection browser.
- 4:21SPEAKER_00: So it's a really good browser.
- 4:22SPEAKER_00: You should check it out.
- 4:23SPEAKER_00: You can look up the SNPs with the RSIDs and see what the results are.
- 4:29SPEAKER_00: In terms of the author list, I don't know everybody here.
- 4:31SPEAKER_00: But I will tell you, Ali Akbar, I don't know him well, but we talked probably for an hour at ASHG.
- 4:39SPEAKER_00: I know his work.
- 4:40SPEAKER_00: I know David Reich.
- 4:42SPEAKER_00: You're moderately well.
- 4:44SPEAKER_00: I've corresponded with Swapan Malik, Nadine Rowland.
- 4:48SPEAKER_00: I don't know well, but I know her name, Ron Panassi.
- 4:50SPEAKER_00: He has a lot of DNA.
- 4:53SPEAKER_00: He's an archaeologist or a genetic anthropologist with a lot of DNA.
- 4:57SPEAKER_00: Eric Lander is a big deal.
- 4:59SPEAKER_00: I don't know why he's on this paper, honestly.
- 5:02SPEAKER_00: Eric Lander is, as some of you know, kind of founder of the Broad Institute, probably the biggest –
- 5:11SPEAKER_00: Mover and Shaker in American Genomics,
- 5:13SPEAKER_00: aside from Jim Watson,
- 5:15SPEAKER_00: maybe,
- 5:15SPEAKER_00: and Francis Collins and Craig Venter.
- 5:18SPEAKER_00: Of those, he's up there, but he's not as known by the public.
- 5:22SPEAKER_00: He's extremely smart.
- 5:23SPEAKER_00: My former boss,
- 5:25SPEAKER_00: Spencer Wells,
- 5:25SPEAKER_00: would tell me that when he was in the Lewontin Lab,
- 5:28SPEAKER_00: Richard Lewontin's lab,
- 5:30SPEAKER_00: Sometimes Lander is a math student,
- 5:33SPEAKER_00: and he came to give a talk or give presentations,
- 5:38SPEAKER_00: and Stephen Jay Gould would be there,
- 5:40SPEAKER_00: and Lewontin would be there.
- 5:43SPEAKER_00: Lander was there.
- 5:43SPEAKER_00: It was very interesting times.
- 5:45SPEAKER_00: And then, of course, Alkes Price is on this paper.
- 5:50SPEAKER_00: Alkes Price,
- 5:50SPEAKER_00: I don't know him,
- 5:51SPEAKER_00: obviously,
- 5:51SPEAKER_00: but I don't want to say that he's like von Neumann,
- 5:56SPEAKER_00: but his students and people that work with him,
- 5:59SPEAKER_00: And when you go to a presentation at a conference, it's like he's von Neumann.
- 6:06SPEAKER_00: His statistical mathematical brain is just not from this world.
- 6:11SPEAKER_00: So the fact that he's on this paper –
- 6:15SPEAKER_00: I don't want to say it reassures me, but it does reassure me.
- 6:17SPEAKER_00: I don't know most of the others by reputation.
- 6:19SPEAKER_00: I mean Nick Patterson has been on this podcast.
- 6:21SPEAKER_00: Nick is a genius.
- 6:22SPEAKER_00: He does a lot of the math, creation of new statistics, Alyssa Mitnick.
- 6:27SPEAKER_00: She's done some good work with the Reich Lab before.
- 6:32SPEAKER_00: So the other people I don't know aside from Ali.
- 6:36SPEAKER_00: But it's actually a relatively short list of authors.
- 6:40SPEAKER_00: That's why I'm actually going over this.
- 6:42SPEAKER_00: It's a short list of authors.
- 6:44SPEAKER_00: I thought there would be a lot more here.
- 6:47SPEAKER_00: But if – so for example,
- 6:49SPEAKER_00: if Alki's price went through everything,
- 6:52SPEAKER_00: I don't think that there are massive technical errors.
- 6:57SPEAKER_00: I'm very skeptical that there would be.
- 7:00SPEAKER_00: And Nick Patterson, David Reich, Ali, none of them are slouches.
- 7:04SPEAKER_00: So this is going to be a nature of science, probably nature if I had to bet.
- 7:09SPEAKER_00: But in any case,
- 7:10SPEAKER_00: there's a massive supplementary info,
- 7:12SPEAKER_00: which is like a book in and of itself that I have not read candidly.
- 7:16SPEAKER_00: And I will at some point.
- 7:21SPEAKER_00: But I'm going to go through this paper.
- 7:23SPEAKER_00: and talk about what they discovered because everyone's curious and talking about it.
- 7:27SPEAKER_00: You know, it's the IQ, the behavioral stuff, but really there's a lot more here.
- 7:31SPEAKER_00: So I'm going to read the abstract really quickly,
- 7:34SPEAKER_00: and then I'll just go through,
- 7:36SPEAKER_00: and I'm going to give some preface about selection adaptation and these sorts of studies.
- 7:40SPEAKER_00: All right.
- 7:41SPEAKER_00: We present a method for detecting evidence of natural selection in ancient DNA time
- 7:46SPEAKER_00: series data that leverages an opportunity not utilized in previous scans,
- 7:50SPEAKER_00: testing for a consistent trend in allele frequency change over time.
- 7:54SPEAKER_00: By applying this to 8,433 West Eurasians
- 7:58SPEAKER_00: who lived over the past 14,000 years,
- 8:01SPEAKER_00: and 6,510 contemporary people,
- 8:03SPEAKER_00: we find an order of magnitude more genome-wide significant signals than previous studies.
- 8:09SPEAKER_00: 347 independent loci with 99% probability of selection.
- 8:15SPEAKER_00: Previous work showed that the classic hard sweeps driving advantageous mutations to
- 8:20SPEAKER_00: fixation have been rare over the broad span of human evolution.
- 8:24SPEAKER_00: But in the last 10 millennia, many hundreds of alleles
- 8:28SPEAKER_00: have been affected by strong directional selection.
- 8:31SPEAKER_00: Discoveries include an increase from 0% to 20% in 4,000 years for the major risk
- 8:37SPEAKER_00: factor for celiac disease HLA-DQB1,
- 8:41SPEAKER_00: a rise from 0% to 8% in 6,000 years of blood type B,
- 8:46SPEAKER_00: and fluctuating selection in TYK2,
- 8:50SPEAKER_00: which tuberculosis risk allele,
- 8:52SPEAKER_00: rising from 2% to 9% from 5,000 to 3,000 years ago before dropping to 3%.
- 8:57SPEAKER_00: We identify instances of coordinated selection on alleles affecting the same trait.
- 9:03SPEAKER_00: With the polygenic score today predictive of body fat percentage decreasing by
- 9:07SPEAKER_00: around a standard deviation over 10 millennia,
- 9:09SPEAKER_00: consistent with the thrifty gene hypothesis that a genetic predisposition to store
- 9:13SPEAKER_00: energy during food scarcity becomes disadvantaged after farming.
- 9:17SPEAKER_00: We also identify selection for a combination of alleles that are today associated
- 9:20SPEAKER_00: with lighter skin,
- 9:22SPEAKER_00: lower risk for schizophrenia and bipolar disease,
- 9:24SPEAKER_00: slower health decline,
- 9:26SPEAKER_00: and increased measures related to cognitive performance,
- 9:29SPEAKER_00: scores on intelligence tests,
- 9:30SPEAKER_00: household income,
- 9:31SPEAKER_00: and years of schooling.
- 9:32SPEAKER_00: These traits are measured in modern industrialized society,
- 9:36SPEAKER_00: so what phenotypes were adoptive in the past is unclear.
- 9:40SPEAKER_00: We estimated selection coefficients at 9.9 million variants,
- 9:44SPEAKER_00: enabling study of how Darwinian forces couple to allelic effects and shape the
- 9:49SPEAKER_00: genetic architecture of complex traits.
- 9:52SPEAKER_00: Okay.
- 9:53SPEAKER_00: There was a lot of terms in there.
- 9:55SPEAKER_00: I'm sure some of you are confused.
- 9:58SPEAKER_00: Just going through really quickly.
- 10:01SPEAKER_00: I'm seeing a hard sweep.
- 10:04SPEAKER_00: So a hard sweep,
- 10:05SPEAKER_00: just so you guys know,
- 10:06SPEAKER_00: and I'm going to talk about this again,
- 10:07SPEAKER_00: is selection at a specific position in the genome of a single mutation,
- 10:12SPEAKER_00: and then it rises in frequency,
- 10:15SPEAKER_00: and that's what's driving the adaptation.
- 10:18SPEAKER_00: Strong directional selection, positive or negative.
- 10:21SPEAKER_00: You probably understand that.
- 10:22SPEAKER_00: Fluctuating selection is frequency dependence going up and down, that sort of thing.
- 10:28SPEAKER_00: And then the allusion to – these traits are measured in modern industrialized society,
- 10:34SPEAKER_00: so what the phenotypes were adaptive in the past is unclear.
- 10:37SPEAKER_00: Well, an example would just be sickle cell in African-Americans.
- 10:40SPEAKER_00: Obviously, we're not farmers in Africa, or they're not farmers in Africa exposed to malaria.
- 10:46SPEAKER_00: and so it's not adaptive today, but it's an echo of adaptation in the past.
- 10:50SPEAKER_00: You know, my family's from Bangladesh.
- 10:51SPEAKER_00: There seems to be a cholera adaptation.
- 10:53SPEAKER_00: I haven't checked if I have it, but if I do, it doesn't really matter in the United States, right?
- 10:57SPEAKER_00: So a lot of these things are the ghosts of adaptation past,
- 11:01SPEAKER_00: but it actually shows,
- 11:02SPEAKER_00: if you look empirically at the results,
- 11:04SPEAKER_00: and I'll go through some of the empirical results,
- 11:06SPEAKER_00: a lot of the adaptations and selection signatures are pretty,
- 11:11SPEAKER_00: are continuing into the present pretty strongly.
- 11:14SPEAKER_00: But first, I want to talk, like, just more broadly about selection and adaptation.
- 11:20SPEAKER_00: And, you know, there's an allusion to Darwinian forces and allelic effects.
- 11:24SPEAKER_00: And I think this is the allusion to the fact that this is obviously in a Darwinian
- 11:28SPEAKER_00: evolutionary biological framework.
- 11:30SPEAKER_00: But the focus here is on genetic signatures,
- 11:33SPEAKER_00: genetic dynamics,
- 11:35SPEAKER_00: using genes as the tracer,
- 11:37SPEAKER_00: tracker,
- 11:37SPEAKER_00: or the currency for evolutionary processes.
- 11:41SPEAKER_00: So, yeah.
- 11:42SPEAKER_00: As I'm assuming you know, selection is necessary for adaptation.
- 11:47SPEAKER_00: So when you go back to Charles Darwin and his theory of evolution through
- 11:51SPEAKER_00: adaptation via natural selection,
- 11:54SPEAKER_00: basically what you have is a situation where adaptation needs heritable variation
- 12:00SPEAKER_00: and then selection correlated with that heritable variation to affect a change in
- 12:06SPEAKER_00: the population.
- 12:07SPEAKER_00: So if there is selection in the population but selection is operating on a trait
- 12:11SPEAKER_00: that's not heritable,
- 12:13SPEAKER_00: then adaptation is not going to happen because the offspring are not going to
- 12:16SPEAKER_00: inherit the trait.
- 12:18SPEAKER_00: And if the selection is – if you have selection but –
- 12:28SPEAKER_00: Yes, if you have selection on a non-heritable trait, it's irrelevant.
- 12:31SPEAKER_00: And if you have the trait and it's highly heritable but there's no selection,
- 12:35SPEAKER_00: obviously adaptation is not going to happen.
- 12:36SPEAKER_00: You have to have both of them, right?
- 12:38SPEAKER_00: And so this is why biologists,
- 12:41SPEAKER_00: evolutionary biologists go into genetics and look at genetics and have Drosophila
- 12:45SPEAKER_00: and other –
- 12:46SPEAKER_00: other sorts of model organisms to study evolution because you use genetics to
- 12:51SPEAKER_00: explore the patterns of evolution because evolution occurs through inheritance
- 12:56SPEAKER_00: generation to generation.
- 12:57SPEAKER_00: It is driven by allele frequency changes.
- 13:00SPEAKER_00: And alleles is just an old generic word that means genetic variant or genetic marker.
- 13:07SPEAKER_00: It basically – today often means SNP,
- 13:10SPEAKER_00: SNP,
- 13:11SPEAKER_00: single nucleotide polymorphism,
- 13:12SPEAKER_00: but we didn't know that DNA was a substrate of inheritance for a long time.
- 13:16SPEAKER_00: So all of the new things and the new terms actually apply to the old terms,
- 13:22SPEAKER_00: but the mapping is a little weird.
- 13:24SPEAKER_00: But an allele just means basically an inheritable cell.
- 13:27SPEAKER_00: It's like a variant that is a variation of a gene,
- 13:32SPEAKER_00: and a gene in the old-fashioned way would be just a discrete unit of inheritance.
- 13:38SPEAKER_00: Today,
- 13:38SPEAKER_00: when we think of a gene,
- 13:39SPEAKER_00: we think of a start codon and a stop codon as biochemical reality,
- 13:42SPEAKER_00: but actually somewhat different than the reality of how genes were initially
- 13:46SPEAKER_00: conceptualized 120-some years ago when genetics was rediscovered after Mendel's
- 13:52SPEAKER_00: period in the 1860s and 70s,
- 13:53SPEAKER_00: right?
- 13:58SPEAKER_00: So you have this situation where you're trying to bridge adaptation of Darwinian
- 14:03SPEAKER_00: forces to genetics,
- 14:05SPEAKER_00: and genetics is your window.
- 14:06SPEAKER_00: It's your tracer.
- 14:07SPEAKER_00: It is – yeah, it's basically a map of evolutionary process and adaptation.
- 14:14SPEAKER_00: But the mapping is not always so easy.
- 14:17SPEAKER_00: And a lot of times you have signatures of selection.
- 14:20SPEAKER_00: You know selection is going on in parts of the genome.
- 14:22SPEAKER_00: But mapping it back to what the pressures of the external world were is not always easy.
- 14:28SPEAKER_00: It is easy if you have Drosophila, if you're breeding flies, fruit flies.
- 14:32SPEAKER_00: But we're interested in humans.
- 14:34SPEAKER_00: We're interested in other organisms.
- 14:35SPEAKER_00: We're interested in the ecosystem.
- 14:37SPEAKER_00: That is much harder.
- 14:38SPEAKER_00: So we have –
- 14:40SPEAKER_00: a lot of data now when it comes to molecular genomic methods but fitting that data
- 14:44SPEAKER_00: to darwinian adaptive stories is very very difficult and i think you'll notice that
- 14:48SPEAKER_00: they were i feel like they're pretty cautious and didn't overemphasize that
- 14:52SPEAKER_00: probably because the methods are pretty intense what they had to do and i'll get to
- 14:57SPEAKER_00: that i also want to say uh you know selection can be non-genetic um
- 15:04SPEAKER_00: So the Cultural Revolution people,
- 15:06SPEAKER_00: Joe Henrich,
- 15:08SPEAKER_00: Robert Boyd,
- 15:09SPEAKER_00: Pete Richardson,
- 15:12SPEAKER_00: Marcus Feldman,
- 15:14SPEAKER_00: the late LL Cavalli Forza – I'm trying not to leave anybody out so that people are
- 15:19SPEAKER_00: going to get annoyed.
- 15:20SPEAKER_00: But – and –
- 15:22SPEAKER_00: Anyhow, Alex Massoudi did a podcast like three years ago now, I think, three summers ago.
- 15:27SPEAKER_00: Was it three summers ago?
- 15:28SPEAKER_00: On cultural evolution.
- 15:29SPEAKER_00: You can listen to that.
- 15:30SPEAKER_00: So evolution can occur through culture.
- 15:33SPEAKER_00: This is the dual inheritance model.
- 15:35SPEAKER_00: And obviously that has nothing to do with genetics.
- 15:37SPEAKER_00: So you have adaptation on a cultural level.
- 15:42SPEAKER_00: You can imagine that.
- 15:43SPEAKER_00: It's a little more difficult to test and select for because there's no genes there.
- 15:48SPEAKER_00: There's memes, and memes are not – no one's really discovered genes.
- 15:52SPEAKER_00: you know,
- 15:53SPEAKER_00: a way to track a meme,
- 15:55SPEAKER_00: if that makes sense,
- 15:56SPEAKER_00: in the same way that we can test a gene,
- 15:58SPEAKER_00: because a gene is a physical entity.
- 16:01SPEAKER_00: It's something that's amenable to technological inquiry,
- 16:03SPEAKER_00: which is really what's transformed genomics over the last,
- 16:06SPEAKER_00: you know,
- 16:07SPEAKER_00: 25 years.
- 16:08SPEAKER_00: So in terms of detecting selection on the molecular level,
- 16:12SPEAKER_00: this started to become a thing,
- 16:13SPEAKER_00: well,
- 16:14SPEAKER_00: 1960s to 1970s.
- 16:14SPEAKER_00: This is the period of the neutral theory, and also...
- 16:22SPEAKER_00: Molecular evolution.
- 16:24SPEAKER_00: And molecular evolution just basically means DNA is a molecule and how is this molecule evolving.
- 16:29SPEAKER_00: Also,
- 16:29SPEAKER_00: there was some early work with other types of variation that's correlated with the DNA,
- 16:35SPEAKER_00: alizimes and protein variation,
- 16:38SPEAKER_00: amino acid variation.
- 16:39SPEAKER_00: I don't want to get into all the different types,
- 16:41SPEAKER_00: but really people mostly focus on DNA now because that's really the important one.
- 16:45SPEAKER_00: That's the ultimate root of a lot of the variation.
- 16:49SPEAKER_00: There's forms of detecting adaptation and selection that are between species or
- 16:54SPEAKER_00: between really diverged lineages.
- 16:56SPEAKER_00: That started out very early.
- 16:58SPEAKER_00: A lot of this comes out of the Drosophila world.
- 17:00SPEAKER_00: So some of you might know DNDS.
- 17:02SPEAKER_00: So DNDS is a statistic that's looking for selection within the genome
- 17:08SPEAKER_00: So you have a gene.
- 17:09SPEAKER_00: So this is going to be in Drosophila usually.
- 17:11SPEAKER_00: And you have two types of Drosophila.
- 17:14SPEAKER_00: And maybe one – I don't know.
- 17:18SPEAKER_00: One can digest like some sort of fruit very well.
- 17:21SPEAKER_00: The other cannot.
- 17:23SPEAKER_00: Well, I mean what's that?
- 17:23SPEAKER_00: That difference is heritable, right?
- 17:25SPEAKER_00: So you know that the flies diverged whatever generations ago or thousands of years
- 17:31SPEAKER_00: ago or hundreds of years ago.
- 17:32SPEAKER_00: It depends.
- 17:32SPEAKER_00: Sometimes they're –
- 17:34SPEAKER_00: Sometimes they're within the same species.
- 17:35SPEAKER_00: But the point is, okay, they have a heritable difference.
- 17:38SPEAKER_00: The differences are usually fixed, which means that they're totally different between the two species.
- 17:43SPEAKER_00: And then you do things like DNDS,
- 17:46SPEAKER_00: which is looking for evolution and,
- 17:51SPEAKER_00: like,
- 17:51SPEAKER_00: changes across the two lineages of non-synonymous positions,
- 17:56SPEAKER_00: right?
- 17:59SPEAKER_00: Okay, I got to go back here.
- 18:02SPEAKER_00: Within the DNA,
- 18:03SPEAKER_00: it's organized into codons,
- 18:06SPEAKER_00: and basically there are some mutations that have an effect on the protein codon and
- 18:12SPEAKER_00: some mutations that don't.
- 18:13SPEAKER_00: The ones that don't are synonymous.
- 18:15SPEAKER_00: The ones that do are non-synonymous.
- 18:17SPEAKER_00: Basically,
- 18:18SPEAKER_00: what you want to do is the non-synonymous ones that mess up proteins or change them somehow,
- 18:22SPEAKER_00: you want to look how fast they're mutating in relation to how fast the ones that
- 18:26SPEAKER_00: are synonymous are mutating.
- 18:28SPEAKER_00: Since the ones that are synonymous or mutating are neutral, they have no up or down effect.
- 18:33SPEAKER_00: They're controlled by drift.
- 18:35SPEAKER_00: The ones that are non-synonymous are usually very, very constrained.
- 18:39SPEAKER_00: Because you don't want them to mutate.
- 18:40SPEAKER_00: You don't want to break the protein.
- 18:41SPEAKER_00: But sometimes the mutation is good in very rare cases.
- 18:44SPEAKER_00: Then you can have positive selection.
- 18:46SPEAKER_00: So one population maybe is adapting to a new fruit.
- 18:49SPEAKER_00: There's positive selection.
- 18:50SPEAKER_00: And so there's a non-synonymous mutation that goes from zero to one,
- 18:55SPEAKER_00: whereas the other population,
- 18:56SPEAKER_00: it stays at zero.
- 18:57SPEAKER_00: It's constrained.
- 18:58SPEAKER_00: And so you compare the non-synonymous positions.
- 19:00SPEAKER_00: Oh, they're different.
- 19:01SPEAKER_00: Well, I mean –
- 19:03SPEAKER_00: Is that region of the genome – what is the variation of the synonymous positions?
- 19:07SPEAKER_00: Oh, this is like way bigger than the variation of the synonymous positions.
- 19:11SPEAKER_00: I kind of used an obvious example, but it's a little bit more subtle than that.
- 19:14SPEAKER_00: Anyway, that's one way you can look at selection across species, right, in the molecular –
- 19:20SPEAKER_00: You know, lens.
- 19:22SPEAKER_00: Another way is called Tahima's D.
- 19:25SPEAKER_00: And,
- 19:26SPEAKER_00: you know,
- 19:26SPEAKER_00: basically a lot of the Tahima's D and these sorts of methods rely on comparing
- 19:30SPEAKER_00: types of variation in the genome.
- 19:33SPEAKER_00: And so,
- 19:34SPEAKER_00: for example,
- 19:35SPEAKER_00: heterozygosity,
- 19:37SPEAKER_00: which is whether,
- 19:38SPEAKER_00: you know,
- 19:38SPEAKER_00: any given position,
- 19:39SPEAKER_00: you have two copies of the gene.
- 19:41SPEAKER_00: any given position of the two copies,
- 19:44SPEAKER_00: the two copies match or don't match,
- 19:45SPEAKER_00: and then compare it to the overall diversity of the genome and look at the ratio.
- 19:52SPEAKER_00: And that tells you whether the population is increasing,
- 19:54SPEAKER_00: decreasing,
- 19:55SPEAKER_00: subject to selection,
- 19:56SPEAKER_00: not subject to selection.
- 19:59SPEAKER_00: It's a little hard to explain why, but a visualization would help.
- 20:02SPEAKER_00: But
- 20:02SPEAKER_00: Basically, these are some old forms of detecting selection.
- 20:06SPEAKER_00: I mean old forms in terms of – they go back decades.
- 20:09SPEAKER_00: They're pretty straightforward.
- 20:10SPEAKER_00: They're usually looking at a single locus or region of the genome,
- 20:15SPEAKER_00: and they're looking at species-level intraspecific differences,
- 20:19SPEAKER_00: looking for differences between species.
- 20:21SPEAKER_00: I'm going to be talking about a totally different thing within species types,
- 20:26SPEAKER_00: obviously within human beings,
- 20:29SPEAKER_00: within our species over the last 14,000 years.
- 20:31SPEAKER_00: So some of these methods,
- 20:32SPEAKER_00: they don't work very well because non-synonymous,
- 20:36SPEAKER_00: synonymous,
- 20:37SPEAKER_00: all of these things,
- 20:38SPEAKER_00: there hasn't been that much time for a lot of mutations to build up,
- 20:42SPEAKER_00: et cetera,
- 20:43SPEAKER_00: et cetera.
- 20:43SPEAKER_00: It's just like the scaling is off.
- 20:46SPEAKER_00: It's very little time.
- 20:48SPEAKER_00: Between the separation, between the lineages.
- 20:50SPEAKER_00: So you need to have different types of selection, right?
- 20:52SPEAKER_00: Or different types of molecular tests.
- 20:54SPEAKER_00: And I'm going to get to that,
- 20:55SPEAKER_00: but first I want to talk about the different types of selection just so that you
- 21:00SPEAKER_00: have this in your head.
- 21:01SPEAKER_00: Positive selection is really straightforward.
- 21:03SPEAKER_00: You have to go zero to one.
- 21:04SPEAKER_00: It's positively selected.
- 21:05SPEAKER_00: It's good.
- 21:05SPEAKER_00: So zero would be like there's a new mutation, and then it goes to 100%, right?
- 21:11SPEAKER_00: Negative selection is the opposite, obviously.
- 21:14SPEAKER_00: It goes from a high frequency and it goes down.
- 21:17SPEAKER_00: Usually it's like you're moving in a new environment or something has changed.
- 21:20SPEAKER_00: So there's something called background selection, which is often – it's related to negative selection.
- 21:26SPEAKER_00: And this is the kind of thing that happens when you get a new mutation that causes
- 21:29SPEAKER_00: a slight bad effect,
- 21:31SPEAKER_00: and it's kind of slowly purged out of the genome.
- 21:34SPEAKER_00: So there's constant background negative selection going on.
- 21:39SPEAKER_00: And that affects some of these statistics that people are generating.
- 21:42SPEAKER_00: You have over-dominance.
- 21:44SPEAKER_00: This is really straightforward, and frankly, I don't think it's that important.
- 21:47SPEAKER_00: But sickle cell is an example of over-dominance.
- 21:51SPEAKER_00: And over-dominance is selection for the heterozygote often.
- 21:54SPEAKER_00: So basically sickle cell, if you have a mutated copy and a normal copy, you're fitter than the other two.
- 21:59SPEAKER_00: The issue with over-dominance is if you do the math –
- 22:03SPEAKER_00: You can't have too many over-dominant positions if you have too many people with
- 22:06SPEAKER_00: too bad copies because independent segregation is a segregation load.
- 22:11SPEAKER_00: Just think about it.
- 22:11SPEAKER_00: Like if you had like,
- 22:13SPEAKER_00: I don't know,
- 22:15SPEAKER_00: 20 over-dominant traits,
- 22:17SPEAKER_00: there's going to be a minority of them where you're messed up probably.
- 22:22SPEAKER_00: And it starts to build up over time.
- 22:25SPEAKER_00: So over-dominance is a real thing.
- 22:26SPEAKER_00: It happens, but I don't think it's super common.
- 22:29SPEAKER_00: This is a much bigger deal, frequency-dependent selection and stabilizing selection.
- 22:33SPEAKER_00: They're related.
- 22:34SPEAKER_00: Frequency-dependent is describing more of the dynamics,
- 22:38SPEAKER_00: while stabilizing selection just means that the trait value,
- 22:40SPEAKER_00: you're going for a midpoint trait value.
- 22:42SPEAKER_00: So imagine you don't want to be too tall, you don't want to be too short.
- 22:45SPEAKER_00: Stabilizing selection keeps you in the middle, kind of, you know,
- 22:49SPEAKER_00: And there's just random genetic drift and segregation,
- 22:53SPEAKER_00: variation,
- 22:54SPEAKER_00: keeps a little variation around the middle.
- 22:56SPEAKER_00: Height, intelligence, maybe those are under balancing selection, stabilizing selection.
- 23:02SPEAKER_00: And often they can be driven by frequency dependence, but not always.
- 23:08SPEAKER_00: But frequency dependence is – there's two types.
- 23:10SPEAKER_00: There's positive and negative.
- 23:11SPEAKER_00: Positive frequency dependence is kind of very similar to positive frequency.
- 23:16SPEAKER_00: positive selection because it just means that the higher your frequency, the more you're selected.
- 23:24SPEAKER_00: Arguably,
- 23:25SPEAKER_00: selection on a recessive trait could be positive frequency dependent because when
- 23:28SPEAKER_00: it's at a really low frequency,
- 23:30SPEAKER_00: the gene isn't expressed very often,
- 23:32SPEAKER_00: so the selection is weak.
- 23:33SPEAKER_00: As it increases in frequency,
- 23:35SPEAKER_00: the gene is expressed more and more as a trait that's selected,
- 23:39SPEAKER_00: and so selection should become stronger at that position.
- 23:44SPEAKER_00: Right.
- 23:44SPEAKER_00: So that's kind of a form of positive frequency dependence.
- 23:46SPEAKER_00: Negative frequency dependence is a lot of the immunological variations where as it
- 23:52SPEAKER_00: increases in frequency,
- 23:53SPEAKER_00: bacterial or viral adaptations occur.
- 23:56SPEAKER_00: And so it's fitness drops.
- 23:58SPEAKER_00: In these cases, if you have a rare mutation, that's often better because there's nothing adapting to it.
- 24:04SPEAKER_00: So these are situations where these regions of the genome are highly diverse.
- 24:08SPEAKER_00: And they're diverse because there's negative frequency dependent selection where
- 24:12SPEAKER_00: when they get really low in frequency,
- 24:14SPEAKER_00: they get extremely adaptive.
- 24:15SPEAKER_00: And then when they get higher in frequency, their adaptiveness drops again.
- 24:18SPEAKER_00: And so they'll often be cycling of some sort.
- 24:21SPEAKER_00: The HLA loci, human leukocyte antigen, these are very, very diverse.
- 24:27SPEAKER_00: And they show up in this paper.
- 24:30SPEAKER_00: I mean, no surprise.
- 24:31SPEAKER_00: I mean, no surprise at all.
- 24:34SPEAKER_00: I want to mention something also relevant to the molecular genetic evolution aspect,
- 24:39SPEAKER_00: hard versus soft sweep.
- 24:41SPEAKER_00: So hard sweep, which is mentioned in the paper, is really like classic.
- 24:45SPEAKER_00: There's a mutation, there's a single mutation, and that just drags like its whole region of the genome.
- 24:50SPEAKER_00: And, you know, it's easy to like see its effect in the genome because it's this one single mutation.
- 24:55SPEAKER_00: It's
- 24:56SPEAKER_00: destabilizing a region of the genome, creating deviations from all sorts of statistics that jumps up.
- 25:01SPEAKER_00: Soft sweeps are more difficult to detect because it's a situation where you might
- 25:06SPEAKER_00: have a bunch of different mutations that are floating around in the population,
- 25:09SPEAKER_00: and all of a sudden they get selected,
- 25:11SPEAKER_00: and so all of them increase in frequency at the same time,
- 25:14SPEAKER_00: but obviously they start interfering with each other,
- 25:17SPEAKER_00: so the signature of selection in that region is going to be weaker because it's
- 25:22SPEAKER_00: different mutations
- 25:25SPEAKER_00: Different variants increasing in frequency, right?
- 25:28SPEAKER_00: So for example,
- 25:30SPEAKER_00: yeah,
- 25:30SPEAKER_00: imagine a situation where 90% of the frequency is one allele,
- 25:38SPEAKER_00: the major allele,
- 25:39SPEAKER_00: and then you have like,
- 25:40SPEAKER_00: I don't know,
- 25:40SPEAKER_00: like 10 different minor alleles at the 10%.
- 25:43SPEAKER_00: What if those 10 different minor alleles have like they break the function of the
- 25:48SPEAKER_00: trait and all of a sudden that's beneficial?
- 25:52SPEAKER_00: What's going to happen is the 90% is going to start to drop.
- 25:54SPEAKER_00: through negative selection against it,
- 25:56SPEAKER_00: and those 10 different minor alleles will start to all rise up,
- 26:00SPEAKER_00: and that's going to be a sweep upward.
- 26:01SPEAKER_00: That's going to be a soft sweep because those 10 are going to be competing with
- 26:03SPEAKER_00: each other,
- 26:04SPEAKER_00: depending on how good they are maybe.
- 26:05SPEAKER_00: But if they're the same, they're just all going to rise up at the same time, right?
- 26:09SPEAKER_00: And so that's the kind of thing that probably is actually kind of similar to what
- 26:14SPEAKER_00: happens with stabilizing selection on quantitative traits.
- 26:19SPEAKER_00: So think about height.
- 26:21SPEAKER_00: There are lots and lots,
- 26:22SPEAKER_00: like hundreds,
- 26:22SPEAKER_00: thousands of genetic positions that control height,
- 26:26SPEAKER_00: and there might be selection for tall or short in some populations.
- 26:30SPEAKER_00: Who knows?
- 26:31SPEAKER_00: Well, in that case, it's not selecting for a single variant.
- 26:34SPEAKER_00: It's selecting for whatever is there within the population, within the individual.
- 26:38SPEAKER_00: And so a bunch of different variants are going to change in frequency together, and it's not like –
- 26:44SPEAKER_00: Technically, like a soft sweep because it's not going from zero to something.
- 26:47SPEAKER_00: It's stabilizing.
- 26:47SPEAKER_00: But this dynamic is common where you have polygenic, many genetic positions subject to selection.
- 26:54SPEAKER_00: Harder to detect,
- 26:56SPEAKER_00: obviously,
- 26:56SPEAKER_00: than the hard sweep from a single mutation that's like a positive directional selection,
- 27:00SPEAKER_00: right?
- 27:00SPEAKER_00: Yeah.
- 27:02SPEAKER_00: I talked to you about DNDs and TGMSD earlier.
- 27:06SPEAKER_00: You know, within species, there's different types of ways to detect selection.
- 27:11SPEAKER_00: So when the genomic revolution happened, we had the whole genome.
- 27:15SPEAKER_00: One thing that was really popular was using haplotype structure.
- 27:18SPEAKER_00: And haplotype structure just basically means the haplotype is basically a sequence
- 27:23SPEAKER_00: of variants in the genome that are correlated together because they were
- 27:26SPEAKER_00: co-inherited together often,
- 27:28SPEAKER_00: you know,
- 27:28SPEAKER_00: with an individual.
- 27:29SPEAKER_00: Because remember, your genome...
- 27:31SPEAKER_00: Your genome is – abstractly, it's organized into DNA, 3 billion base pairs, but it's two copies.
- 27:37SPEAKER_00: And those two copies are physically organized into 20 – well, 23 –
- 27:43SPEAKER_00: Okay, 22 autosomes, X chromosome, and the Y, depending on if you're male or female.
- 27:47SPEAKER_00: Two Xs if you're female, okay?
- 27:49SPEAKER_00: That's how it's packaged,
- 27:50SPEAKER_00: and they're organized that way,
- 27:52SPEAKER_00: and the recombination swaps them around that way,
- 27:54SPEAKER_00: but they're linear.
- 27:55SPEAKER_00: They're linear, and so there's two sets of your genome, really, right?
- 28:00SPEAKER_00: And those two sets have mutations and variations that are linked together.
- 28:05SPEAKER_00: If you...
- 28:07SPEAKER_00: For example,
- 28:07SPEAKER_00: my children,
- 28:08SPEAKER_00: they have two sets of the genome,
- 28:11SPEAKER_00: and one set actually has a lot of variation from South Asians.
- 28:15SPEAKER_00: That's me.
- 28:16SPEAKER_00: The other has a set of two, and then the other genome is Northern European, their mom.
- 28:23SPEAKER_00: And so the haplotypes are very distinct, are going to be very distinct when recombination happens.
- 28:29SPEAKER_00: They have offspring because the segments are going to be very distinct from each
- 28:33SPEAKER_00: other because they have different clusters.
- 28:36SPEAKER_00: The genes have different clusters of variations that define their haplotypes.
- 28:40SPEAKER_00: OK, so haplotype is a line.
- 28:42SPEAKER_00: It's a segment of the genome that defines correlated,
- 28:45SPEAKER_00: correlated markers,
- 28:46SPEAKER_00: correlated variants,
- 28:47SPEAKER_00: usually SNPs.
- 28:49SPEAKER_00: And when you have selection,
- 28:51SPEAKER_00: when you have a hard sweep,
- 28:53SPEAKER_00: the haplotype structure gets totally disrupted from normal expectation because
- 28:59SPEAKER_00: usually the correlations across the segment of a genome is pretty small because
- 29:03SPEAKER_00: recombination broke apart the associations.
- 29:06SPEAKER_00: So your genome is swapped back and forth 20 to 30 times or 20 to 40 times per generation.
- 29:12SPEAKER_00: Over the generations, the correlations disappear.
- 29:14SPEAKER_00: But then you have a hard sweep.
- 29:16SPEAKER_00: And then what happens is the whole region around that mutation that's good –
- 29:25SPEAKER_00: hitchhikes along with it.
- 29:27SPEAKER_00: Eventually, you know, over the generations, the segment will get shorter and shorter.
- 29:34SPEAKER_00: That's pulled along because recombination breaks apart and swaps, right?
- 29:39SPEAKER_00: But you have, like, for example, lactase persistence, LCT, that gene is very, very, has a huge haplotype.
- 29:48SPEAKER_00: And that's because it was an extremely strong signature of selection.
- 29:52SPEAKER_00: That was very recent.
- 29:53SPEAKER_00: So those are the two conditions.