Hello and welcome back to Introduction to Genetics and Evolution. We've been talking about the Hardy-Weinberg Equilibrium and how, under certain conditions and only under those conditions, you can calculate or estimate expected genotype frequencies from observed allele frequencies. The opposite is always true. You can always know allele frequencies from genotype frequencies. But you cannot always infer genotype frequencies from allele frequencies. Now in the last example I showed you, the last video we saw a case where there were fewer heterozygotes, fewer of the Aa individuals, observed relative to what was expected from Hardy-Weinberg. Now why might that be? Well this is gonna be the first of many possible deviations from Hardy-Weinberg that we'll discuss. And it could be what's referred to often as the Wahlund Effect. Well let's look at a little bit of real data. Here's some real data from a Navajo population at the MN blood groups. MN, just like big A and little a, there's two alleles, M and N. There's three possible genotypes. MM, MN, and NN. So let's take a look at what we see when we look at the Navajo of this particular group. Well, we can do the same tests for Hardy-Weinberg we've done before. We figure out the total number of individuals, 361 in this case. Get the genotype frequencies, the true observed genotype frequencies. From them, say all of these plus half of these, there are allele frequencies. So the frequency for big M is 0.971, for big N it's 0.083. Take this squared, .841. 2 p q, so two times this times this. 0.152. 0.83 squared is 0.007. Now we see, this population's not absolutely at Hardy-Weinberg, but it's very close, it's very close to the Hardy Weinberg predicted frequencies. So let's look at another population now, let's look at these Aborigine. Now again if we look at the MM blood group in these individuals, let's follow the same procedure. I won't go through all the steps but if you want some practice you can pause the slide, take these first set of numbers and go through it yourself. We come back to again, a set of genotype frequencies that are predicted, and they're very close to those observed. 0.031 is very close to 0.030. 0.293 versus 0.296. .676 versus .674. It's within .003 for all of these things of the expected genotype frequencies. Now let me ask you a funny question. What if you were an alien from another planet, and you came down and you grabbed a bunch of Aborigine, and you grabbed a bunch of Navaho, separately, and you put them all together into the loading bay of your spaceship. You have this mixed population. So let's say it's exactly the same individuals who are genotyped, and we saw the Aboregenie are at Hardy-Weinberg. The Navajo were at Hardy-Weinberg. What happens when we put all these together? Well, we get these allele frequencies for M and N, but we get something that's deviating rather dramatically from Hardy-Weinberg expectations of the genotype frequencies. Look at that. This is dramatically different, this is dramatically different, and this one as well. You notice especially, and this is what I want to point out in particular, we expect almost half the individuals to be heterozygote. We observe only about a quarter of individual heterozygote. So there's a dramatic deficit of heterozygotes in the observed relative to the expected. That's like the last example from the last video. Now why might we see this? Why do we see this deviation? We have a Hardy-Weinberg population and another Hardy-Weinberg population, why is it when we put them together It's not a Hardy-Weinberg? What assumption have we deviated from? Well one big assumption we deviated from the list I showed you earlier, was the assumption of random mating. The idea is that any two individuals are as likely to breed as any other two individuals. Remember from the first video from the series, that gametes just floating all around. It's not like that there because the Navajo lady is not as likely to breed with an Aborigine as she is with another Navajo. So imagine that in one population big A is abundant. Then big A's are gonna be very likely to encounter other big A's. In little population, little As are very abundant. Little a's are gonna be very likely to encounter other little A's. But big A's and little a's are very unlikely during counter each other that's why you see this deficit of heterozygous. And then in this regard [NOISE] the Hardy Weinberg assumption was violated. That Hardy Weinberg assumption was not rejected within the Navajo, or within the Aborigine, but it deviates from this combined population and this results from nonrandom mating. And importantly, this will very typically result in having too few heterozygotes. We expected about half. We observed about a quarter. This pattern is referred to as the Wahlund effect. This is when you sample a cross population. So the populations within each population may have random mating. But when you sample across or between populations you get an under-representation of heterozygotes relative to Hardy Weinberg. So this is a way for potentially identifying different populations. You can see how much of this deviation you see. We'll use that, actually, in a subsequent video for calculations that are referred to as FST. But let me ask you a different question first. Why does it matter? Why does it matter if something's a Hardy Weinberg or not? Well in fact, the first step in genome-wide association studies for genetic diseases, or any trait, is Is or should be to test for Hardy Weinberg. Now why is that? Well, actually geno wide association studies assume Hardy-Weinberg is true or assume that you're very very close to it. basically you're assuming that there's linkage to this equilibrium. linkage to this equilibrium caused by close proximity between marker alleles and disease causing alleles. Remember that fundamental purpose of all genetic mappings to see an association between genotype and phenotype. And we're hoping this association is from close proximity or lack of recombination. So imagine that you see something like, where 20% of individuals with AA genotypes have a disease. And 5% on individuals with aa genotype have a disease. Then we're assuming there's an association between the AA allele, or the AA genotype, or the A marker gene more broadly, and the disease. But just being in different populations also causes linkage disequilibrium. Let me give you an extreme example to illustrate this point. Let's imagine that in Population 1, every individual is AA. Okay? Just that simple. Imagine in Population 2, every individual is aa. Now let's say that a disease is a little more abundant in Population 1 than in Population 2. Okay? Would you say that AA individuals are more likely to have a disease than aa individuals? The answer is yes, you would say this because just the way it laid it out. Now this is actually a fake LD between disease and the gene. Because the disease may not be on the same chromosome as the A gene in particular, and in fact, the disease may not even be genetic. Lets say, for example, in population one every eats a lot, in population two everybody has very good weight. You may see obesity is much more abundant in population one than population two, but it may have nothing to do with your genotype at the A gene. So its really important that you have true Hardy-Weinberg in doing these genome wide association studies. Otherwise the associations you see may have nothing to do with the genotypes you're observing. And the disease in fact not even be genetic at all. Now punchline is if there are allele frequency differences between populations at a SNP, which is very often true. Let's say the SNP is being used as a marker. And if disease incidence differences exists between the two populations you're studying, which again, very often true. Sometimes for genetic reasons, sometimes for not, but it may not have anything to do with the particular marker or anything near that marker you're looking at. Then a genome-wide association study will erroneously make it seem that a gene near the SNP is causing or contributing to the disease. Now if you test for Hardy-Weinberg then you can avoid this error because you can identify if you're looking at one interbreeding population or not. Your hope is that the population is at Hardy-Weinberg or is very, very, very close to being at Hardy-Weinberg. And then if you see an association you know it's not this weird bias. Now, although it's very important to test for Hardy–Weinberg, this is often not done. Here are excerpts from two studies from not too long ago. This is from the American Journal of Epidemiology 2006. The exclusion of studies in which Hardy–Weinberg was violated changed the conclusions and changed the statistical significance of gene-disease associations. That's scary. Think about it. Millions of dollars go into finding these gene-disease associations. We really need to be carefull and know that we're doing them right. Here's something from the European Journal Human Genetics in 2005. Testing and reporting for Hardy–Weinberg equilibrium is often neglected and deviations are rarely admitted in the published reports. So this is a really big deal. There's other issues about interpreting the deviations for Hardy-Weinburg. Let me show you an example here. So this is a real example where a Hardy-Weinburg test was done, but interpreted incorrectly. This is raw data from a 2000 study of BRCA2 variants. These are from newborn males from a hospital in the United Kingdom. Just as a little test here, I want you to look, or I want you to do the math for this and figure out how close this is to Hardy-Weinburg expectations. Do you see a particular kind of deviation? So try that out. Well hope that wasn't too hard, let me go ahead and show you the answers, these are the numbers you should have come up with, so these are the two. Genotype frequencies. These are the true allele frequencies. These are the Hardy Weinberg expected genotype frequencies. What we see here is our expected frequency of the heterozygote is 0.4, our observed was .36. So there is, or there at least seems to be, some slight deviation for Hardy Weinberg, and in this direction of too few heterozygotes. This was, by the way, statistically significant, too. Interestingly, how did the authors interpret this? The authors of the study interpreted it as that the Aa individuals are less healthy than AA or aa. They postulated that maybe there was some disease or there was some problem associated with Aa individuals. In fact, there is a much simpler explanation. That we're looking at newborn males in a hospital in the United Kingdom. It's quite likely, imagine this is a hospital in a place like London. It's quite likely that in a place like London, there's a lot of subdivision of the population. That people of say, Indian descent are probably more likely to have kids with others of Indian descent. People from the Far East may be more likely to have kids with people from the Far East. People who, who are of European decent are probably more likely to have kids with others of European decent. Yes there are cases where people will have kids with people from other ethnic groups. But by having this tendency there, overall the population, which undoubtedly exists, you will get exactly this pattern. It's basically the simpler explanation than the Aa individuals are less healthy, is that there's a little bit of a Wahlund effect there, but that wasn't considered. It probably didn't take our pop gen class. Now let me close with a little ironic tidbit. This is a quote from Hardy's 1940 book. This is when he was about 62. His book is called A Mathematician's Apology. I definitely recommend it. It's a very interesting read about the elegance and beauty of math. He said, I have never done anything useful. No discovery of mine has made or is likely to make, directly or indirectly, for good or ill, the least difference to the amenity of the world. I would like to say very strongly, he was very wrong on this. This Hardy-Weinburg idea, which he helped bring about and he helped popularize, really has made a huge impact. We're continuing to see it now even literally more than a hundred years after the original publications of the Hardy-Weinburg work. We still see these applications for things like genome-wide association studies. The kinds of things he never would have imagined. So this is really cool stuff, even though this was written when he was 62, and he was lamenting the waning of his mathematical ability, he really did a lot for the world. As did, of course, Weinberg and Castle, thank you.
