recent selection

The first of the papers describing results from the 1000 Genomes project has been released today in Nature [1].

This is "big project" genomics news. Like many announcements of this kind, it represents more of a public relations milestone than actual scientific advance. Some of the project data have been publicly available for a while -- the 1000 Genomes and HapMap projects have to their great benefit been based on the strategy of immediate data release. The new paper and its supplements include many summary statistics and report on new genetic variants that have been found -- there's a lot of information here. But most of the interesting science is just getting started. A paper like this really represents the opening of a race to use the new data for innovative research.

Here in my lab, we are exploring the ways that whole genome sequencing can change our study of human population history. A large part of this is our work on recent selection, of course ("Why human evolution accelerated"). Whole-genome sequencing is not essential to finding many recently selected regions of the genome, but it will help enormously in narrowing down the actual functional changes that affected fitness in past populations.

Whole-genome sequencing will rapidly improve our ability to resolve the population history of Pleistocene humans. For older events -- going back to the origins of Homo -- whole-genome sequencing will give us samples of genealogies from across the genome. We will be able to resolve some very ancient episodes of population mixture, and we have a chance of testing what kinds of events accompanied the rise of our genus. Even for events of the Late Pleistocene and Holocene, for which haplotypes of SNP markers can be useful without resequencing, whole-genome sequencing can be tremendously valuable. Reconstructing haplotypes from diploid genotypes requires us to make some assumptions about the demography of the population, which may be exactly what we are trying to discover. A sample of genomes sequenced at high read coverage will free us from some of those assumptions. It's really exciting stuff for an anthropologist.

All those are reasons why the data will be useful for us in the long term. But at the moment, the data are not nearly so rich. The current paper reports:

1. Whole-genome sequencing at 42x coverage of six individuals, one three-person family trio from Utah, and one family trio from Nigeria.

2. Low-coverage (2x-6x) sequencing of 59 Yoruba, 60 Utah residents, 30 Chinese and 30 Japanese individuals. These are a subsample of the original HapMap samples.

3. Sequencing at 50x coverage of 8140 exons in 697 individuals. These are a subset of the HapMap v.3 population samples, including Yoruba, Luhya, Utah, Tuscan, Japanese and Chinese samples. These exons come from 906 genes targeted "randomly".

It's pretty far from a thousand genomes, and even farther from the stated goal of 2400 genomes. The low-coverage genomes are not sufficient to call genotypes across most of the genome. This is a persistent problem with "whole-genome" sequencing projects so far. A person's whole genome is mostly diploid -- two copies of most everything. Recently, we've seen several "whole-genome" sequences where each base is given a consensus value. SNP variants may be called against other people's genomes, but rarely is there sufficient coverage to call SNPs within the individual. There are exceptions -- a handful of public whole genomes are at high coverage. The exon sequencing here should be enough to call SNPs in these functional regions with great confidence. The family trios also should have enough to call SNPs. So some of these will be our first chance to do actual population genetics on diploid genome-wide sequence data.

One important piece of analysis in the paper is the confirmation of a low rate of de novo mutations in the children of the family trios. I discussed a result last spring that came to a very low rate of per-site mutation ("A low human mutation rate may throw everything out of whack"). The rate in that paper was 1.1 x 10^-8 per site per generation. The current paper comes to a rate between 1.0 and 1.2 x 10^-8. I have some more written on this issue and I'll integrate the new finding and post it later in the week. This aspect of the study is pretty important to our understanding of human evolution.

The paper makes an interesting distinction between "accessible" and "inaccessible" portions of the genome -- accessibility meaning ease of mapping and aligning sequence reads:

Accurate identification of genetic variation depends on alignment of the sequence data to the correct genomic location. We restricted most variant calling to the ‘accessible genome’, defined as that portion of the reference sequence that remains after excluding regions with many ambiguously placed reads or unexpectedly high or low numbers of aligned reads (Supplementary Information). This approach balances the need to reduce incorrect alignments and false-positive detection of variants against maximizing the proportion of the genome that can be interrogated.

For the low-coverage analysis, the accessible genome contains approximately 85% of the reference sequence and 93% of the coding sequences. Over 99% of sites genotyped in the second generation haplotype map (HapMap II)4 are included. Of inaccessible sites, over 97% are annotated as high-copy repeats or segmental duplications. However, only one-quarter of previously discovered repeats and segmental duplications were inaccessible

It's an interesting decision -- just focus and report on the majority of the genome where alignment is easier.

The paper discusses selection briefly. There's not much new here other than the identification of candidate causal variants for some selected haplotypes.

First, it provides a more comprehensive catalogue of fixed differences between populations, of which there are very few: two between CEU and CHB+JPT (including the A111T missense variant in SLC24A5 (ref. 38) contributing to light skin colour), four between CEU and YRI (including the −46 GATA box null mutation upstream of DARC39, the Duffy O allele leading to Plasmodium vivax malaria resistance) and 72 between CHB+JPT and YRI (including 24 around the exocyst complex component gene EXOC6B); see Supplementary Table 7 for a complete list. Second, it provides new candidates for selected variants, genes and pathways. For example, we identified 139 non-synonymous variants showing large allele frequency differences (at least 0.8) between populations (Supplementary Table 8), including at least two genes involved in meiotic recombination—FANCA (ninth most extreme non-synonymous SNP in CEU versus CHB+JPT) and TEX15 (thirteenth most extreme non-synonymous SNP in CEU versus YRI, and twenty-sixth most extreme non-synonymous SNP in CHB+JPT versus YRI). Because we are finding almost all common variants in each population, these lists should contain the vast majority of the near fixed differences among these populations. Finally, it improves the fine mapping of selective sweeps (Supplementary Fig. 14) and analysis of the dynamics of location adaptation. For example, we find that the signal of population differentiation around high Fst genic SNPs drops by half within, on average, less than 0.05 cM (typically 30–50 kb; Fig. 5d). Furthermore, 51% of such variants are polymorphic in both populations. These observations indicate that much local adaptation has occurred by selection acting on existing variation rather than new mutation.

This last point is not especially demonstrated by the new sequencing data. What we are looking at is few complete sweeps, but that's expected even if all the selected variants were novel mutations -- there just hasn't been time to fix many variants. It remains to be shown the extent to which standing variants are involved in this selection, partial sweeps of new mutations, or parallel adaptations ("Spatial dispersal, parallel adaptation, and the 'Stooge effect'"). We'll probably see a lot more interesting work on recent selection coming out of the new data.

Science has a companion paper to the Nature data summary, focusing on copy number variation and gene duplications. I will review that one separately.

UPDATE (2010-10-27): Dienekes pulls out an interesting passage about the Y chromosome sequences, which in at least one case recover many markers separating haplogroups once thought to be much closer to each other. Not sure what to make of that yet.

References

Peter Ralph and Graham Coop have an interesting paper in the current Genetics, titled, "Parallel Adaptation: One or Many Waves of Advance of an Advantageous Allele?" [1]

Fisher [2] famously considered the case in which an advantageous allele is dispersing through a spatially dispersed population, showing that the dispersal forms a "wave of advance". This work was the foundation for a lot of progress in understanding spatial dynamics of organisms.

As I discussed in 2008 ("Overstating the obvious"), one of the consequences of the Fisher wave model for human evolution is that advantageous alleles will spread very slowly through the population. During the course of the Holocene, a strongly selected mutation might move only across a radius of a thousand or so kilometers. That provides one explanation for why new advantageous alleles haven't spread very far beyond their points of origin -- they just haven't had time yet.

Another reason why an allele might not have spread widely is interference from other alleles with similar effects. I mentioned this process last year ("Spatial variation and near-fixed selected alleles"):

Greg Cochran and I have been discussing this idea for some time. We call it the "Stooge effect". Think of the Three Stooges all trying to run through a door at the same time and getting stuck in the middle. That's what these genes are doing -- all of them are competing to respond to selection, but each is slowed by the presence of the others.

Ralph and Coop have cleverly combined the "Stooge effect" phenomenon with spatial dispersal. They suppose a case in which two separate advantageous mutations arise in different geographic locations, each affecting the same trait. Each begins to spread independently as a Fisher wave of advance. What happens when they meet?

As they show, the dynamics in this case give rise to a static equilibrium -- once the "waves of advance" meet, they stop moving, forming a stable boundary. A new favorable mutation makes headway only so long as it has no equally favorable mutation to compete against.

I like the way they used both analytical approaches and simulations to come to this outcome. The appearance of stable boundaries in a reaction-diffusion system has long been known (demonstrated first by Alan Turing, actually!). But to my knowledge, no one has considered this specific case from an analytical perspective.

The Fisher equation is not all that simple for most students to work with. If you become familiar with the equation, you will notice the key aspect is that it has two separate components -- a logistic (or reaction) component representing the increase in frequency at a single point in space, and a diffusion component representing the dispersal across space.

The muscle of the dispersal process comes from the logistic component. Without the intrinsic growth of the selected allele, the dispersal of individuals along the boundary would not carry many copies of the selected allele into new geographic areas. If the local selective advantage dies, the wave of advance rapidly stalls. A static equilibrium arises, with the frequency of the selected allele forming a cline that correlates with the local selection pressure.

Ralph and Coop's model approximates this case, in a dynamical sense. Each new selected mutation forms an increasing zone in which the selective advantage of other mutations is zero. When those other mutations encounter this zone, they form a stable cline. The cline is stable in the short term, but the diffusion component still disperses copies of an allele; they just lack the muscle to continue their deterministic expansion.

The most interesting simulations by Ralph and Coop show the two-dimensional case, in which the stable boundaries emerge in a "tesselation" pattern.

Figure 6 from Ralph and Coop (2010), showing "tesselations" in 2-d simulations of waves of advance.

The lower three panes in the figure show the stability of the boundaries between the selected alleles. They proceed to fixation locally, but their dispersal stops where they come into contact with other adaptive alleles. Over the very long term, the population will mix -- the diffusion process will slowly carry all these alleles throughout the species' range. Look at the process after a million generations and the entire zone will be gray. But this dispersal occurs at the neutral rate, where the diffusion term is the only factor driving the dispersal.

What about humans?

My graduate student Zach Throckmorton and I have been working in this area for a while now. One of the things that impresses us is the way that much more interesting dynamics can emerge when you alter the assumptions. I learned some of this stuff by talking to Frank Livingstone, who gave a lot of thought to these issues of spatial dispersal and selection as applied to malaria resistance alleles.

In particular, Frank thought about the case where one allele has a slightly larger advantage than another. In some contexts, this allows the "better" allele to overtake and swamp the expansion of the "weaker" (but nonetheless adaptive) one. In others, the two come to a near standstill, one displacing the other only very gradually. Much depends on the timing of the two mutations and the local conditions controlling their initial dispersal.

Ralph and Coop briefly consider this case in their paper, noting that the difference in fitness advantage of two alleles will allow one to advance into the range of the other, albeit at a slower rate. In humans, we may be seeing a smaller subset of cases, where one or more of the alleles have not yet established a wavefront. In these cases, the arrival of another wave can disrupt the spatial pattern of the rarer allele. The diploid case gives rise to the possibility of more complex epistases. Well-defined boundaries between selected alleles are rare, and where they occur (as may be the case with HbC and HbS in Africa), many have focused on negative epistasis as an explanation.

Also, alleles are unlikely to substitute perfectly for each other. In many cases, they may work synergistically -- individuals carrying two selected alleles that affect the same function may outperform those carrying only one such allele. At some point, new selected mutations may start to have diminishing returns, even on a trait like skin pigmentation where dozens of alleles may have been selected in widespread human populations. So the current distribution may to some extent be "frozen", but by a more complicated dynamic than the simple intersection of waves of advance.

As Coop and colleagues showed last year [3], and we discussed in 2007 [4], there are really only few genes that have approached local fixation in recent human evolution. The current spatial pattern of recently selected alleles doesn't look like a tesselation with many alleles near local fixation. Over most of the Old World, it looks like populations have a very large number of very new alleles, far from fixation, and few up over 70 percent in frequency.

So the specific scenario in this paper by itself probably does not explain the overall empirical pattern in humans. But if we consider the current pattern as a transient, approximating the early stages of dispersal for many selected alleles, we may not be terribly far off the mark.

Mutation-limited evolution

This is a long dense paper and there's a lot in it. One further aspect of the paper that I think is essential is the way that Ralph and Coop reiterate the basic point that more people means more mutations. In their case, they focus on population density over space (population number, when you multiply them) as a constraint on the number of possible adaptive mutations. They apply this idea as a hypothesis to account for parallel adaptations that may have emerged in recent human evolution.

Multiple mutational origins are likely if the characteristic length is shorter than the physical dimensions of the region. Eurasia measures >8000 km across, and so Table 1 suggests that multiple origins at a single base pair are very unlikely at the lower population density. On the other hand, if the mutational target is large, then multiple origins are likely at low densities, while at high densities independent origins are ubiquitous. The complementary cases of (rho = 2, µ = 10^–8) and (rho = 0.002, µ = 10^–5) give identical characteristic lengths of 3000 km, although the timescale on which the mutations spread differs. Thus for these two parameter combinations we can expect a few mutations to dominate within continents and for multiple mutations to be common in a population spread across an area the size of Eurasia. Obviously these calculations are very crude, as population densities vary through space and time, and dispersal across continents is not simply a function of geographic distance and individual dispersal. Nevertheless, these calculations suggest that it is plausible that for adaptive traits with reasonable mutational targets (e.g., a change anywhere within a gene or pathway) even low population densities can lead to parallel adaptation across an area the size of Eurasia, and higher densities almost certainly will.

We note that as human population densities have increased dramatically over time, so too has the probability of parallel adaptation. It is interesting therefore to note that a number of recent human adaptations (e.g., sickle cell alleles) involve repeated changes at very small mutational targets in relatively small geographic areas, while older adaptations from single changes (e.g., skin pigmentation) are more broadly spread.

They are describing a scenario in which small human populations would have been mutation-limited -- that is, the number of new mutations is small, making it unlikely that adaptive mutations will happen in any given generation. In such populations, the rate of adaptation is limited by the availability of new mutations. In an extreme -- in the very small effective sizes of Pleistocene human populations -- the rate of adaptation may be extremely slow and regional populations may come to differ at many weakly selected loci, which spread very slowly.

As the population grows, strongly adaptive mutations become more and more likely to happen somewhere in the species' range. Yet they are still relatively rare -- meaning that they have an opportunity to spread fairly far before encountering another equally strongly selected mutation affecting the same trait.

This process can give rise to very large differences on a continental scale, even when the selection pressures in different regions do not differ. In humans, the dispersal of selected alleles across space may have been significantly accelerated by actual dispersals of populations. It is not a mere coincidence that very widespread alleles in Eurasia also tend to be much older than 20,000 years old -- long-distance dispersals prior to that time had a higher chance of leaving a lasting influence on subsequent populations.

But as the population gets bigger and bigger, parallel mutations are more and more likely to happen. As Ralph and Coop point out, at the extreme of large population size and likely mutations, you shouldn't see any new mutations emerging and spreading over very large areas. Any of these mutations would be very likely to encounter other new mutations that do the same thing.

Is this likely in humans? Clearly some mutations have happened recurrently. Making a broken gene is easy -- there's a large mutational target, since a large fraction of nonsynonymous substitutions might do the job. So if there's a net selective advantage to breaking a gene, we ought to see that happen recurrently in human populations.

In contrast, if the mutational target is very small, then mutations will still be rare even in a very large population. If only one base change can have an adaptive effect, that precise change will happen less than once in 10⁹ births (remember that not just any mutation at a site, but some particular mutation is what we may need). If a rare duplication or gene conversion is the necessary change, then it may be much rarer.

Looking across the last few million years, when human population numbers were much smaller than the Holocene, we can be pretty sure that some aspects of our evolution were mutation-limited. The changes that took hold in our ancestors were the ones that happened, and that survived the winnowing of genetic drift. Many changes that would have been adaptive didn't happen in our ancestors. They just weren't lucky enough.

But some of those changes would still be adaptive now, if we could get them. And we have had much larger numbers in the last 10,000 years. Homo erectus needed these mutations, but we only now are seeing them selected in the human population.

Malaria adaptation

Hemoglobinopathies are among the cases of easy mutations -- where breaking a gene is adaptive. It's not just any broken version of alpha- or beta-globin that does the job, though. The hemoglobin needs to be impaired in certain ways to impede the parasites while maintaining blood function. This provides many of the classic cases of human adaptation, and Ralph and Coop turn to this system for examples of parallel adaptation:

The sickle cell allele HbS at the β-globin gene in humans provides a particularly interesting case of putative parallel adaptation. The HbS allele (β6 Glu-Val) has been driven to intermediate frequencies by selection within the past 10,000 years due to increased resistance to malaria of heterozygotes for the allele (HALDANE 1949; ALLISON 1954; CURRAT et al. 2002; KWIATKOWSKI 2005). The HbS allele is present on at least four major distinct haplotypes in Africa, each at intermediate frequency within a different geographic region; the haplotypes are named after the population sample where they were first discovered (Central African Republic, Senegal, Benin, and Cameroon). This is consistent with multiple origins of this single-base-pair change. Note that a distinct, malaria resistance allele, HbC (β6 Glu-Lys), has also arisen in Africa at the same codon as the HbS allele (TRABUCHET et al. 1991; AGARWAL et al. 2000; WOOD et al. 2005a), increasing our confidence that the mutational input was high enough to allow multiple types to arise. However, FLINT et al. (1998) thought the hypothesis of multiple new mutations arising at a single base pair was extremely unlikely and proposed that it was more likely that gene conversion had spread a single mutation across multiple haplotypes.

The theory we have developed can be used to assess the plausibility of the multiple mutational origins of the sickle cell allele, by exhibiting parameter combinations that yield characteristic lengths consistent with the separation of the sample locations. [Recall that the wave of advance, and thus also our model, works in the case of heterozygote advantage (ARONSON and WEINBERGER 1975).] The different HbS haplotypes co-occur within a few thousand kilometers of each other (see Table 5 of FLINT et al. 1998) (noting that these locations are unlikely to reflect the geographic mutational origins, and mutations will have been spread by large population movements). As the HbS changes occur at a single base pair, the mutation rate would have been 10^–8, and we take an s = 0.05 (as in CURRAT et al. 2002). If human dispersal at that time was well approximated by a Gaussian kernel with sigma = 100 km, then a characteristic length of 1000 km would require an effective density of individuals of rho = 25 km^–2, while if sigma = 10 km, then we would require only rho = 2.5 km^–2. This latter set of parameters does not seem unrealistic, considering our knowledge of population density and dispersal parameters, so our model suggests that the hypothesis of multiple origins is not unreasonable.

I think they've got the basic idea correct here, but there are some additional details to consider. The distribution of HbE is not quite so easy to understand if parallel mutations are really so likely, and of course there is the negative epistasis of different alleles (and the thalassemias) which impacts their dispersal ability when they become moderately common. The dynamic may be of similar form to the one described here, but boundaries between alleles may be reinforced by the fitness costs of carrying multiple ones.

This situation raises the issue of path dependence. Some mutations have "first mover" advantages. Once they are common, other adaptive mutations may still occur -- even mutations that are better from the standpoint of fitness -- but be lost or grow very slowly because their net fitness advantage over the common mutant is slight. Where HbE is common, new HbS alleles are unlikely to invade quickly. Where HbS is common, new HbE mutants are similarly unlikely to invade -- even though HbE has a higher fitness.

Network effects among genes may also dominate the spatial dynamics. HbS spread most widely in the context of populations that were already Duffy null, and in which G6PD deficiency was rapidly increasing. The first conditioned the parasite environment -- P. vivax had a strong disadvantage in Duffy null populations, P. falciparum made up most of the parasite load. G6PD deficiency should have impacted the relative advantage of HbS, more and more as it became more common. Those are two loci among many that alter malaria dynamics in Africa compared to South and Southeast Asia.

Conclusions

There is much more to say about this paper -- it's 22 journal pages. But I think I've given an impression of what's there and how the ideas may impact our interpretation of recent human evolution. Many of the central concepts were presaged by earlier work in 2007 and 2008, as reviewed here on the blog. The new analytical and simulation work, I really like.

Hopefully we can get out some shorter papers that will focus on aspects of these problems as applied to humans. A message that comes across very clearly in our work and this new paper is that different time periods in our evolutionary history must have had very different selection dynamics. Pleistocene humans were not only in a different ecology than us, they experienced a radically lower potential for adaptation.

References

Malaria in humans is caused by one of five different species of Plasmodium parasites. The deadliest of these is P. falciparum, especially within Africa where native resistance to P. vivax is high. Where the vivax parasites seem to have been around for at least tens of thousands of years, P. falciparum in many ways looks relatively young. Its comparative lack of genetic variation suggests either a recent origin from some other primate species, or an intense bottleneck or selective sweep affecting the parasite's demography. In either case, the falciparum history seems to indicate that its present widespread distribution is a very recent phenomenon -- possibly within the last 5000 years.

Because P. falciparum is phenotypically similar to the major chimpanzee malaria parasite, P. reichenowi, most scientists have assumed that we got falciparum malaria from chimpanzees. But in a new report, Weimin Liu and colleagues [1] have surveyed parasite variation in gorillas, bonobos and chimpanzees across Africa, finding that human falciparum parasites all group in with a single small clade of gorilla parasites. The other primates carry many varieties of parasites, with typical individuals being highly heteroplasmic -- that is, carrying several different strains.

From the discussion:

Using single-template amplification strategies and a much larger collection of ape specimens than previously analysed, we show here that wild-living chimpanzees and western gorillas are naturally infected with at least nine Plasmodium species. Among more than 1,100 SGA-derived mitochondrial, apicoplast and nuclear gene sequences from 80 chimpanzee and 55 gorilla samples, we found a total of nine sequences that were related to P. malariae, P. ovale or P. vivax (Supplementary Table 5). All others grouped within one of six chimpanzee- or gorilla-specific lineages representing distinct Plasmodium species, three of which had not previously been described. Significantly, all currently available human P. falciparum sequences constitute a single lineage nested within the G1 clade of gorilla parasites. This indicates that human P. falciparum is of gorilla origin, and not of chimpanzee9, 10, 12, bonobo11 or ancient human5 origin, and that all known human strains may have resulted from a single cross-species transmission event. What is still unclear is when gorilla P. falciparum entered the human population and whether present-day ape populations represent a source for recurring human infection. It has been suggested that the limited levels of genetic diversity seen at many loci in human P. falciparum reflect a relatively recent selective sweep8. Our data suggest that this bottleneck or ‘Eve event’ was instead the consequence of cross-species transmission of a gorilla parasite. It is difficult to date this event without having reliable dates with which to calibrate the Plasmodium phylogenetic trees.

What's interesting about the study is the sheer coverage of wild primates, and the application of multiple gene trees, which suggests that this is a recent origin of human parasites instead of introgression and selection of a single gene. I don't know if it makes any difference whether the disease came from gorillas or chimpanzees, but it certainly helps to confirm that it is new and not a long-time coevolution. That explains the burst of recent selection associated with resistance genes, especially within Africa.

References

Bray and colleagues [1] report on genotyping of 471 people of Ashkenazi Jewish descent. This is one of the largest samples of a single human population, and is therefore very interesting for studies of population history and recent natural selection.

There's a lot in the paper. One of the key findings in the paper is that the Ashkenazi population doesn't look bottlenecked -- in fact, it looks outbred compared to Europeans generally. The paper also documents a high amount of admixture with non-Ashkenazi Europeans, ranging from 35% to 55%. Figuring out the actual history of the population -- when and where its ancestors lived and how they interacted with other people -- is beyond the scope of this kind of analysis. But I expect that somebody can put together a really compelling historical account using these data.

I turned quickly to the issue of selection. They are able to substantiate evidence of positive selection on several disease-causing alleles in the Ashkenazi population, including the Tay-Sachs allele. The lack of evidence for bottlenecks or founder effects pretty much takes away the alternative explanation. Yet they were unable to show statistical evidence of selection on some other disease-causing alleles in Ashkenazi populations:

To explore whether regions of selection in the AJ population included any loci of known Ashkenazi diseases, we examined 21 disease- and cancer-susceptibility loci with known mutations found at higher frequency in the Ashkenazi population. Only 6 of the 21 genes fell in or near (within 500 kb) the top 5% of the AJ iHS windows (Table 2). Among these is the Tay-Sachs disease gene, HEXA, whose selection has been widely debated (4, 5, 14–16) and was found ~400 kb downstream of a window on chromosome 15 identified in the top 1% of the AJ iHS hits. Although none of the SNPs interrogated immediately adjacent to the HEXA locus showed elevated iHS signals, it is possible that the nearby region may contain regulatory elements under selection that affect HEXA expression. Cochran et al. (14) speculated that selection of many of the AJ- prevalent disease loci, especially the lysosomal diseases, conferred an increase in intelligence that was necessary historically for the AJ economic survival. Our data shows evidence of strong selection at or near only six disease loci, including only one out of the four AJ- prevalent lysosomal storage diseases, thus arguing that most AJ disease loci are not under strong positive selection, but rather rose to their current frequency through genetic drift after a bottleneck. However, we cannot exclude the possibility that selection of some AJ disease loci are outside the limits of detection by the extended haplotype tests, which are known to have less power to detect se- lection of lower frequency alleles (38, 41).

It seems to me that this passage probably wasn't written by the same author who showed the lack of evidence for founder effects a few pages before. In this case, the confusion probably comes from the fact that the "detection of positive selection" is actually a refutation of the hypothesis of genetic drift. With a larger sample it will be possible to test the hypothesis with greater power.

Ddisease-causing alleles are at low frequencies currently, making them unlikely to rise to the top percentages of the statistics. It would be interesting to control for current frequency, but I haven't seen a test that uses frequency information in this way.

It's quite remarkable to reflect on the idea that positive selection has now been demonstrated on six disease-causing alleles in the Ashkenazi population. Every one of these is a case of overdominance -- where the heterozygote carrying an allele has some selective advantage, while the homozygote carrying two copies has a disorder. I was having a conversation with a very prominent geneticist a few months ago, who claimed that no case of overdominance in humans had ever been demonstrated except sickle cell. Now, that was obviously false even at the time -- as I pointed out, the many hemoglobinopathies are fairly clear examples. But we've come an awfully long way.

From data like these, we're going to learn a huge amount about low-frequency selected alleles. The Tay-Sachs-causing allele is one of the most common recessive lethal genes in any human population, but like all genes subject to strong selection in homozygotes, it remains rare. Finding selection on these kinds of alleles is very hard unless sample sizes increase to several hundred individuals. Here we are seeing evidence of selection in historic populations -- within the last 2000 years. More will be coming.

References