Joined: 30 Mar 2005 {Posts: 1057 } Location: New Jersey
Posted: Tue 16 Jan 2007 18:13 Post subject: HLA relatedness not necessarily the same as admixture level
I spoke briefly with my mother's friend's husband, who lives in Germany. He is a geneticist who uses HLA. I asked him about genetic distance calculations that determine relatedness between populations, and wondered how an allele or locus known to be under selection would be affected by this. I wasn't certain how the genetic distances are calculated, and if this is similar to determining admixture rate. Apparently, they are not one and the same.
Heinz said that genetic distance calculations don't determine rate of admixture, like we're used to thinking of when using mtDNA and Y chromosomes rates to estimate amount of admixture. Rather, they simply determine relatedness, and so are not really affected by an allele under selection.
Normally, with an adaptive allele, like HbS, calculating the percentage that this allele appears in a population, and then using that as the estimate of the admixture reflected by the allele, will show inflated results, and not show true level of admixture, because it has been expanded through selective pressure.
When calculating HLA genetic distances, the frequencies of the various alleles are compared with eachother to determine relatedness. If an foreign allele enters a population, and is very beneficial, its frequency will become inflated, because the population carrying it will increase, and the population lacking it will decrease. However, an increase in the population carrying the allele naturally translates into an increased relatedness between the foreign population from whence it came and the population it migrated to. HLA genetic distances will reflect this increased relatedness, even though it doesn't mean there is an increased rate of admixture. Therefore, even though its frequency in the new population isn't an accurate reflection of rate of admixture, if one is simply calculating genetic relatedness, as is done with HLA alleles, it accurately reflects this.
Does this make any sense to anyone? Frank?
I see what he's getting at, but never thought in those terms before.
Some Hyperdiffusionists try to use HLAs as proof of precolombian contact. Some responses I have had from geneticists on specific HLA claims.
Prof. Dr. M. Luiza Petzl-Erler wrote:
The HLA allele group A*31 (that is, allele A*310102, the most usual of the A*31 alleles) has a wide distribution and occurs in populations of all continents. The possibility of an "early intercourse between Brazil and Africa" in not supported by the frequencies of this allele in some African and Amerindian populations. In fact, HLA data (various loci / many alleles) support the hypothesis of Asian origin. Even the many HLA alleles seen only in Amerindians of (nowadays) Latin America and nowhere else (notice that A*31 is not one of these alleles) are most easily explained by local origin, involving a set of ancestral alleles brought to America by the Asian founder populations.
In summary: the global distribution of the A*31 allele does not support and early direct contact between African and American populations. The most plausible explanation for the high frequency seen in Amerindians and in some other populations is random genetic drift and natural selection.
For a discussion about the HLA polymorphism in Amerindians (origins of alleles vs. origins of the Amerindians), see:
Parham P, Arnett KL, Adams EJ, Little A-M, Tees K, Barber LD, Marsh SGE, Ohta T, Markow T, Petzl-Erler ML. Episodic evolution and turnover of HLA-B in the indigenous human populations of the Americas. Tissue Antigens 50:219-232, 1997.
Dr. Richard Villems wrote:
HLA is a system which is very likely under very strong natural selection, though we may not necessarily understand the nature of such selection in any particular case. Strong selection can alter allele frequencies very fast - 1000 times faster that fluctuations in neutral genes usually occur. Furthermore, looking to some other and more neutral genetic systems - such as mtDNA - there is not a slightest hint about Pre-Columbian gene flow between Africa (Sahara or sub-Saharan alike) and Americas. One may say that such a flow did not involve women (a reasonable explanation). However - the Y chromosome is a very unreliable marker system here because virtually all Amerind Y chromosome pools of different tribes encompass a considerable, sometimes very large share, of European, and depending on tribe, European and African Y chromosomes. Finally - Amerind tribal gene frequencies are in general very spotty - people believe that this is because of random genetic drift in small isolated populations. Therefore - frequency differences even in an order of magnitude scale are not really surprising.
fwsweet wrote:
This section of chromosome #5 is the weapons factory for our species's eternal war against germs. Its genes produce the lymphocytes and antibodies that defend us. Because of this, it has two features that make it unsuitable for tracking migrations, in my view. First, it evolves and mutates at a phenomenal rate. Nearly every generation carries some new twist in this section of DNA. (You can see why this must be. If it were not so, the germs would have won long ago.) The [Guthrie's] article also mentions other markers (including the cde blood types, other blood proteins, and mtDNA). Judging by those observations, it appears to me that the author may tend to interpret evidence of admixture as evidence of precolumbian transoceanic contact, rather than as merely showing post-colonization mixing. It is not clear how he would tell them apart. Specifically, I have no problem with the sentence, "The surprisingly high frequencies in the Mande, Tigre, and Tuareg samples could be the result of early intercourse between Brazil and Africa." I just do not find it persuasive evidence of precolumbian contact. The mixed Portuguese in colonial Brazil imported huge numbers of mixed slaves from northwest Africa.
Dr. Bernard Ortiz de Montellano wrote:
The use of mtDNA and Y-chromosome uses sections that are not adaptive so that mutation and thus the distributions are due to random occurrences. HLA however depend on what exposure the populations have been exposed to. If different populations are exposed to the same infections, they will end up with similar HLAs even though then may not be related or ever in contact. I'm sure people try to compensate for that, but inherently HLA will be less accurate.
Joined: 30 Mar 2005 {Posts: 1057 } Location: New Jersey
Posted: Fri 19 Jan 2007 17:16 Post subject:
Dr. Bernard Ortiz de Montellano wrote:
If different populations are exposed to the same infections, they will end up with similar HLAs even though then may not be related or ever in contact. I'm sure people try to compensate for that, but inherently HLA will be less accurate.
I'm sure molecular anthropologists who specialize in HLA would have something to say about that! It seems as though there is an ongoing war between those who specialize in using mtDNA and Y-chromosomes and those who use HLA. They seem to be at eachother's throats, constantly trying to discredit one another's methods. I personally believe all forms of molecular anthropology (analysis of HLA, mtDNA, Y-Chr., classical markers) have some validity. Let's face it -- they're all used extensively, and I doubt they would be if they were unreliable. Very often they support eachother.
For example, the HLA study on the Greeks, by Arnaiz-Villena, et al., which is utterly despised by Dienekes Pontikos (a Greek White Nationalist) and Racial Myths (an Italian-American White Nationalist), shows sub-Saharan genetic material in some Greeks. There are eleven alleles at the DRB1 locus that sub-Saharans and Greeks share; some of these are found sporadically elsewhere, mainly in nearby regions, while others are found nowhere else. To me, this is definitive proof of admixture, as I can't fathom so many identical alleles appearing in two populations who have had absolutely no contact with one another. But other studies also show sub-Saharan admixture in Greeks. There is the Benin strain of HbS; an African beta-thalassemia allele; Chromosome 7 markers (cystic fibrosis mutations typical of sub-Saharans, Greeks being the only Europeans that have these); East African mtDNA M1; the Y-Chr. A marker; and autosomal studies by AncestryByDNA show low levels of statistically significant sub-Saharan admixture in Greeks. There is also E-M78, which originated in East Africa and spread to Greece many thousands of years ago; this shows ancient admixture. So, here, the HLA study's findings are corroborated.
The study, also by Arnaiz-Villena, that shows a genetic relatedness between Palestineans and Jews (and was retracted for politically incorrect wording), also is supported by other studies.
I have heard comments made about HLA alleles being subject to selection (and therefore being unreliable), but interestingly, Arnaiz-Villena says the same thing about mtDNA and Y-Chr., and defends HLA:
Arnaiz-Villena, et al., 2002 wrote:
The HLA system has proved to be useful to define ethnic groups, even when its polymorphism was still low. [...] Most HLA-analysed populations show relatedness or unrelatedness according to geography. This is a sign that we are dealing with a good genetic marker. When historical "characteristics" are present, HLA is able to detect them: for example, the A29-B44 haplotype is frequent in northern Spaniards (including Basques), Irish, southern British and western French. Thus, this marker relates all these groups and their history (or prehistory) must have common genetic grounds.
Other molecular markers, like mtDNA and Y Chr. are widely used for this type of research. They are also subject to selection since they are linked to diseases [...url link...], as are a number of HLA alleles.
[...]
Therefore, HLA genes constitute one of the best tools to study relatedness between populations.
Regarding claims of proof via HLA of pre-Columbian African contact with Latin America, I am very skeptical. Although I don't really agree with Petzl-Erler that the globally-present A*310102 allele, despite its much higher than normal frequencies in Amerindians and Africans, isn't suggestive of admixture between these two groups, I would conclude that the admixture is post-Colubmian, and not pre-Columbian. Since it is known that Africans and Amerindians (and Europeans) all mixed in Latin America, I don't know how on earth they can claim the admixture the genes reflect is from before Columbus' time. It's not like this is an allele that existed in Africa before the 1400's and then went extinct before Columbus (which would be more persuasive evidence of such a claim).
Last edited by William on Tue 23 Jan 2007 20:30; edited 1 time in total
Sure HLA's can be secondary support, but the fact still remains that they can be influnced in way that mtdna and y-chromosome just can't. SO when there is HLA but no major mtdna or y-chromosome similarities, I will still be skeptical.
Joined: 30 Mar 2005 {Posts: 1057 } Location: New Jersey
Posted: Tue 23 Jan 2007 17:57 Post subject:
Salsassin wrote:
Sure HLA's can be secondary support, but the fact still remains that they can be influnced in way that mtdna and y-chromosome just can't. SO when there is HLA but no major mtdna or y-chromosome similarities, I will still be skeptical.
But those who specialize in HLA say similar things affect mtDNA and Y-Chr. research. That is why I said the researchers engaging in various types of genetic research are always at war with one another in terms of the validity of their methods. The one camp attacks the other. I personally have quite a bit of faith in the HLA system.
Here is an e-mail I got yesterday from one who specializes in HLA (althouth it primarily concerns a study retracted for political reasons):
William wrote:
Hello, Prof. Arnaiz-Villena:
I have learned much from your HLA studies on populations. I know one (the one on relatedness between Jews and Palestineans) has been retracted solely for political reasons, but a few scientists also said basing conclusions on the DRB1 locus is ill-advised because it is only one locus and is under selection. A few others have said the entire HLA system is under selection. Personally, I feel these supposedly scientific criticisms with regard to the retracted study are merely to support the political withdrawal of the study. To me, it seems impossible that many alleles (even if at only one locus) of one population could match exactly those of another without admixture having occurred. Also, most HLA studies come up with results that are very similar to mtDNA and Y-Chr. studies.
I'm just writing to confirm the validity of DRB1, and HLA in general.
Thanks.
Dr. Arnaiz-Villena wrote:
Thank you for your letter.
You may realize that mtDNA and Y-Chr markers behave (even if called haplotypes ) as a single locus with alleles, and they are also subject to selection since they are linked to diseases.
I must say that DRB1 alleles have stronger discrimination than mt or Y markers.
But the DRB1 critics did only point to DRB1 alleles while in our studies, but genetic distances at this and other loci, quasi-specific (for populations) alleles belonging to other HLA loci, correspondence analysis including DQB1 alleles and extended HLA haplotypes were used to reach conclussions, in additoion to DRB1 alleles, which as you said so many alleles could not appear in two totally unrelated populations.
I don't think much of HLA studies. The fact that they are not random but rather under selective pressure renders them suspect as far as I'm concerned. I also happen to think that the Arnaiz-villena Greek-African paper is quite suspect (and not just because the blogger is Greek) the quote from the textbook and the graph showing how Villena's data is suspect that is posted here is pretty graphic:
I also read the letter to Nature from very respected geneticists about this paper and its faults here
Nature 415, 115 (10 January 2002); doi:10.1038/415115b
Neil Risch from Stanford U., Alberto Piazza from the University of Torino and L. L. Cavalli-Sforza also from Stanford had to say on the interpretations of Arnaiz-Villena based on the HLA-DRB1 marker [1]:
Even a cursory look at the paper's diagrams and trees immediately indicates that the authors make some extraordinary claims. They used a single genetic marker, HLA DRB1, for their analysis to construct a genealogical tree and map of 28 populations from Europe, the Middle East, Africa and Japan. Using results from the analysis of a single marker, particularly one likely to have undergone selection, for the purpose of reconstructing genealogies is unreliable and unacceptable practice in population genetics.
The limitations are made evident by the authors' extraordinary observations that Greeks are very similar to Ethiopians and east Africans but very distant from other south Europeans; and that the Japanese are nearly identical to west and south Africans. It is surprising that the authors were not puzzled by these anomalous results, which contradict history, geography, anthropology and all prior population-genetic studies of these groups. Surely the ordinary process of refereeing would have saved the field from this dispute
I also consider Arnaiz's paper on the Mazatecs to be invalid
Joined: 30 Mar 2005 {Posts: 1057 } Location: New Jersey
Posted: Thu 01 Feb 2007 15:57 Post subject:
Yeah, I've already addressed all of that in the Sticky. I've been over these reports with a fine-toothed comb, and feel my conclusion is accurate.
The article written by the three scientists was not about the Greek study, but about a completely different study on Jewish / Palestinian relatedness, which was retracted for political reasons only and which happened to cross-reference a Greek-sub-Saharan relatedness from the Greek study (I hope Bernard Ortiz de Montellano realizes this); their article claiming the study is scientifically unsound was written simply to "substantiate" the political retraction. Pontikos quotes only part of the article to make the unsuspecting reader think it is referring to the Greek study. Arnaiz-Villena admitted to me in an e-mail that Cavalli-Sforza was pushed into contributing to the article. The scientific criticism is a straw-man, since it is claimed that Arnaiz used only the DRB1 locus, when in fact he also corroborated his data with other loci. If the 3 scientists really were against the use of the DRB1 locus, they'd have complained about all the studies that use this, not just the on on Jewish / Palestinian relatedness. If the genetic community at large were against the use of HLA data, there wouldn't be so many studies done that employ it.
Jobling takes issue with the supposed arbitrariness of one correspondence analysis (the one based on the DRB1 locus). He mentions nothing about the other one based on other loci, about the genetic distance calculations, or about the finding of sub-Saharan alleles in the Greek population. Jobling himself states that Arnaiz knew the population origins (including where the strange alleles came from in Greeks), so he isn't questioning this.
Pontikos is clever by putting all of this together to try to discredit the study on Greeks, but it is dismaying that he is influencing scientists, who should know better, and at least delve a little deeper, and not simply accept distorted data put up by Greek white nationalists who certainly have an axe to grind. My whole point in writing the sticky on Greeks was to clarify and defend Arnaiz, because I feel his results and conclusions are perfectly valid. In fact, the very regions Arnaiz found sub-Saharan HLA alleles in Greeks (the Aegean Islands, the Athens area, and Cyprus) correspond to the regions that Snowden mentions sub-Saharan presences (based on historical accounts and physical evidence, like statuary, etc.).
Furthermore, other genetics studies using different methods have found sub-Saharan material in Greeks. So, there is corroboration.
Bernard Ortiz de Montellano obviously is a believer in mtDNA and Y-chromosome studies only. But, as Arnaiz has said, these can be under selection, too. This just reinforces what I said about scientists using these methods and those using HLA; each thinks his method is better than the other, and never the twain shall meet. It is unfortunate.
Last edited by William on Fri 02 Feb 2007 18:31; edited 1 time in total
Yeah, I've already addressed all of that in the Sticky. I've been over these reports with a fine-toothed comb, and feel my conclusion is accurate.
FWIW, I have also come to the conclusion that the retraction of the A-V paper on Israeli-Palestinian relatedness was grossly political, and that the fraudulent use of the retraction to discredit the unrelated A-V paper on Greece is the work of Greek "racial purity" cranks.
On the other hand, as I have mentioned before, I am not very comfortable with HLA-based phylogeography in general. I have explained why it bothers me elsewhere, but the main problem is that you cannot date the mutations. HLA mutates faster than any other stretch of code, it mutates at irregular intervals, and different stretches mutate at different rates. A lesser problem is that it is adaptive.
"Heinz said that genetic distance calculations don't determine rate of admixture, like we're used to thinking of when using mtDNA and Y chromosomes rates to estimate amount of admixture. Rather, they simply determine relatedness."
Relatedness is all there is, unless you're racial essentialist. Uniparental markers don't show degree of relatness between populations, anyway.
From
[b]Analysis by DNA[/b]
"How is BGA analysis different from mtDNA and Y-chromosomal ancestry analysis?"
"There are several commercially available tests of mtDNA(female contribution) and Y-chromosomal(male contribution) markers, which have been promoted as a means of learning one’s ancestral origins. Although these tests could provide information regarding the provenance of some of a person’s ancestors, they are very limited. For example, one generation ago a person has two ancestors, one mother and one father; five generations ago, a person has 32 ancestors; while 10 generations ago, a person has 1024 ancestors. Ten generations is roughly 250 years and within the time frame of genealogical interest, especially when we are considering the settlement of North America, because they only look at two (2) chromosomes. Y-chromosomal analysis and mtDNA analysis each could only provide information on a very small proportion of a person’s ancestors. Our test relies on sequences throughout your genome, so we can say more about a greater number of your ancestors."
Also:
The distribution of genetic variation within and among human populations has long been an important tool for inferring the evolutionary history of modern humans. Dramatic improvements in genotyping technologies over the past 15 years have facilitated the development of many types of DNA markers. Considerable attention has been devoted to both uniparental and autosomal genetic markers. Because of their lack of recombination, uniparental markersmtDNA and the nonrecombining region of the Y chromosome (e.g., see R. L. Cann et al. 1987; Ingman et al. 2000; Underhill et al. 2000)and their genealogical histories are perhaps easier to study than are recombining markers. Although recombination introduces additional uncertainty regarding the history of any individual autosomal locus, consideration of a large collection of polymorphic loci spread across the genome enables more general inference about demographic history and population relationships than does study of the Y chromosome and mtDNA, loci whose histories may be anomalous when compared with that of an "average" locus in the genome. Studies of autosomal variation that are based on protein polymorphisms, blood groups, restriction-site polymorphisms, and Alu insertions have revealed much about within- and among-population genetic diversity of humans (Cavalli-Sforza et al. 1994; Relethford 2001).
'
FAQ
Search
Register
Log in
