Hello, and welcome back to introduction to genetics and evolution. In the previous video we talked about the evolutionary advantages of recombination. In this video we're going to be looking at something a little bit different. We want to look at how recombination will affect patterns of neutral DNA sequence variation. DNA sequence variation that may not matter in and of itself, but it is actually impacted by say the spread of adapted variants. We'll look at two related concepts, selective sweeps and hitchhiking, in this video. We left off the last video by asking a question, how does recombination affect molecular evolution? And we showed, in the previous video that recombination can help combine good mutations and this is good for the population because you don't have to wait for two beneficial mutations to arise in the same genetic background. It also helps you get rid of bad mutations from a population. It stops this process that we refer to as molars ratchet which is the steady deterioration of fitness in a population. As you accumulate more new mutations, you can't get rid of them. Such that essentially you lose the mutation free class and the population as a whole gets worse and worse and worse. But again the question we want to ask now is, how does recombination affect variation in neutral sequences? Sequences that have no direct effect on fitness. Well the first question we have to ask before we ask that is, how do we quantify molecular variation as a whole? So we have here nine different sets of sequence. And these are potentially from different individuals. So this could be from individual one, two, three, four, five, etc. Well we look at this, and most bases are the same. So the stretch of three Ts is the same in all nine individuals. All right, we're assuming at this point that you only have one copy, we're not looking at the two copies that each individual have, but let's say we randomly selected one copy from each individual. There are three sites that are variable. There's a site variable here, here, and here. They're depicted in green. Is it sufficient to just to say there are three variable sites out of however many there are here, maybe about twenty or so? No, because you notice that they're very different. So here at this first site, we have one individual who has a G but everybody else has an A. So that's an example of a very rare variant of that site. Here we have multiple individuals who have a T, so a little bit less than half and several have an A, and here it's very close to 50/50 and there is four Ts and five Cs. So what we need to do is find some way of capturing these relative abundances in our measure of how much variation there is. So one commonly used metric for measuring the amount of molecular variation in DNA sequences, when you have sequences from many individuals, is referred to as pi. Okay? Well, pi just very simply put is the average number of pairwise mismatches among all the sequences you have. This is analogous the 2pq idea that we used back when we were originally studying Hardy Weinberg. It's kind of like the average number of predicted heterozygotes. So let's look at these three example sequences I have here. Between the first and second sequences, how many mismatches are there? Well we can look very simply, there is one here, one here, one here. Made it very easy by making them red? Well we have three mismatches between the first and second sequence. What about between the second and third? Well, those same three positions also differ between the second and third sequence. So there you go, we have another three. What about between the first and third sequence? How many mismatches are there? Well, the only variable sites, are the red ones. And, here they're the same. Here, they're the same. And, there they're the same. So, in fact, between the first and third sequence, there are 0 mismatches. Now, pi is just the average number of pairwise mismatches. So, we have, in this case, 3 numbers for mismatches, 3, 3, and 0. And we had three pairwise combinations, so we just add these up and divide by the number of pairwise combinations, which in this case was three, we got three different comparisons. 3 + 3 = 6, 6 / 3 = 2. Now it's not very useful just to use this two by itself because it could be two in 10 bases, two in 100 bases, two in a 1000 bases, so typically when people talk about pi, they average out by the number of sites being investigated. In this particular case we looked across a stretch of 20 bases. So pi per site would be two divided by the number of bases, which in this case was 20. So in this case, pi would be 0.1, or simply put, 10%. Now pi is going to be much more affected the more differences you have among sequences, so this is kind of interesting in that if you have a very rare variance, it's not actually going to raise pi all that much within a population. Whereas common variation if say half the individuals have C, and half the individuals have T, you have a lot more opportunities for pairwise mismatches. So pi per site is greater when there are more bases differing among individuals. And pi is not as much affected by very rare variants. Let's apply this and see what's happening in terms of studying neutral sequences and spread of variation. So again, the question is, how do patterns of recombination affect variation in neutral sequences? Now what happens, although these particular sequences, or the variance that are found at these sequences, may be neutral. There may be an occasional beneficial mutation that arises. So what happens when you have an occasional beneficial mutation? How does that affect the neutral variation everywhere else? So sequences are not intrinsically neutral, it just happens to be the variation that's present at those sequences that is neutral. So let's assume a case where we have no recombination whatsoever, we have this variation here at base one is neutral. In this case now we have three Gs, and about six As. At base two, we again have these As and Ts, at base three we have Cs and Ts but we're assuming all this variation is neutral. None of this is affecting fitness. Now, in this base four here, and let's say that it's possible for there to be an adaptive mutation at base four. And let's say here it is. So here this G that has arisen by mutation is adaptive. Well if there's no recombination across this entire stretch, what's going to happen? Well, since this G here is adaptive, it's going to spread. And since there's no recombination, it's basically glued to this whole set of sequence. So as this G spreads, so will this C, so will this A, so will this A. Right, so let's watch what happens over time. The G starts to spread. So G now is across half the population, guess we erased all the variation in that half of the population. Those other three bases. As G spreads completely, we've lost all variation not just at the adaptive site but at these three sites that previously had neutral variation. Okay, notice this though, we had the spread of an adaptive variant and we had a loss of variation at these other three sites. And not only did we lose variation, but we specifically carried this one variant, the C, the A and the A, from those three sites, along with the spreading G. So what we witnessed is called hitchhiking associated with a selective sweep. So, a selective sweep is the spread of an advantageous allele through the population by natural selection. It's what we expected from number four. Right? That fourth base was the adaptive variant. And importantly, the loss of variation associated with the sites near it. So, it's swept out all the variation that was present there before. Hitchhiking is a related concept. This is the spread of nearby alleles along with the advantageous variant because of linkage or because of this lack of recombination. So in this case at SNPs one two and three we had the A A anc C variants that all hitchhiked along with the spread of that adaptive variant at base four. So the sweep there is what got rid of all the old variation and hitchhiking is what happened in terms of these particular three variants. Now this was the case when there was no recombination whatsoever. What would happen if we actually said there was lots of recombination among these sites, that they weren't all glued together as one variant? Let's try that. So now, this is exactly the same slide that we saw before. So what if there was lots of recombination among these sites? Well, we might initially have, this variant right here is going to spread. And it is in the very first generation it is associated with this A, A and C at bases one, two and three. So, okay, it spreads a little bit. But now there's some opportunity. Let's say for example this one has a kid with one of those. There's opportunity for some recombination to happen. So, a heterozygous individual experiences a cross over before making a gamete. So initially, we have this AAC associate with the G but we have this other bit, we have this other haplotype GTT, well what might happen is we might have a crossover just like that which essentially puts this G and C next to a T and G over there, producing this new gamete. You see this? So this is something that was not found in the parentals, right? This is a recombinant, because the G and C were previously together. The T and G were previously together, but this group was not previously with this group, because you see, one of them came from this haplotype, one came from that haplotype. So that get's moved into the egg. We now have a case where there's the G adaptive variance associated with the T and G over there, rather than just with the two As. Note the C there is still with the G because it was a lot closer. So we might have you know more spread. So what's happened in this case? This is a little bit different now. We look here at base one. There's still some variation. We look here at base two, there's still some variation. Base three actually has a little variation, but it lost a lot of it. This one little T slipped in, but mostly you lost a lot of the variation there. So, let's break this up. So, what happened with recombination? While the mutation at site four still spread, that's very important, it's not that selection stops, selection operated just the same as it did before. Recombination allowed SNPs one and two to maintain actually most of their variation, that we still have two abundant alleles at those two SNPs, right? So, the other alleles recombine onto the chromosomes baring the sweeping allele at site four. So, pi was actually not reduced in that left half of the picture that we just saw. SNP 3 had something that was kind of intermediate between the two. At SNP 3 most of the population got one allele. Right we had that one variant that spread but there was a rare one that there that didn't. So most of the chromosomes have the same allele at SNP 3 after the sweep. So there was a loss of variation at SNP 3 but not to zero just lost some of it. And there was one particular allele that became very common by hitchhiking, okay? So in this case, if you go back, let me go back actually a couple of slides. So you go back here, looking at this. There is, if you split this in half, there is a big loss of variation in this side of the picture. All right. Whereas there is basically no loss of variation in this half of the picture. Okay? So pi is greatly reduced in the right half unlike the left, and we did actually have a reduction variation at SNP 3 but we did have hitchhiking still happen as well. So this is pretty cool, and this is the kind of thing that we expect happens as we have the spread of adaptive variants through human populations, or through populations of any species. Let me show you one hypothetical example. It's actually not hypothetical. It's a real example. This is looking at variation at a single nucleotide polymorphism near the EPAS1 gene. Now this affects tolerance to low oxygen. So there were some interesting studies looking at high-altitude Tibetan populations, and sure enough, what it turns out, these are very high-altitude populations. This affects tolerance to low oxygen. There's evidence for a recent sweep and hitchhiking in these Tibetan populations. If you look at a particular SNP, the A variance in Europe is at a frequency of about 0.46, right? Now there is a mutation to tolerance for low oxygen that arose in China. It arose on a chromosome that had this A allele. So in these Tibetan populations, we now see that A SNP frequency is 0.89. So this A SNP hitchhiked along with the adaptive variant at EPAS1 and it's beneficial to have this new variance. So, this is something that's often used a signature of natural selection. Now let me recap three principles here. First, adaptive alleles can sweep, that means that they spread through the population and they eliminate some variation nearby them. The second is that SNP and or marker alleles, you know SNPs are markers, that are near the adaptive allele. By near I mean that they don't recombine away from it. And not necessarily referring to physical distance, I'm more referring to combinational distance. So basically if they have zero or low recombination from the adaptive allele, they will tend to hitchhike, and pi will typically be reduced at those SNPs or in those areas. In contrast, SNPs or marker alleles that are far from an adaptive allele, so, in other words, that there's a lot of recombination between them and the active allele, they do not hitchhike, and pi is not typically reduced. You did not see the evidence of a sweep in those cases. So, this is pretty cool and there's actually exercises to demonstrate this if any of you are interested. This is actually one developed by my laboratory. This is something where you can essentially put this together in your lab. There's a paper on it in the journal Evolution Education & Outreach by Heil et al. Heil is a PhD student of mine. Essentially you start this population with a bunch of white eyed fruit flies. These white-eyed fruit flies are healthy. They're okay. They're doing fine. We introduce a red-eye allele. That's actually the wild type allele. What ends up happening is after several generations you see that most of the flies end up having red eyes because this red-eye allele is advantageous. Now this bottom part shows you a chromosome, so the eye color gene is on the x chromosome and here's the red white eye color gene. Now you can do PCR or ways of assaying two particular genetic markers. One of them's called near. One of them's called far. Near happens to be right next to the eye color gene. Far is very far away. Now let me ask you in closing here, what do you expect? Which of these two market alleles should hitchhike along with the red eye variant? And then two related questions. Do we expect pi to be reduced at the near region? And do we expect pi to be reduced at the far region? So try those out, and then check out the next videos. Thank you for your time.
