Genetic Similarities Within and Between Human Populations
The proportion of human genetic variation due to differences between populations is modest, and individuals from different populations can be genetically more similar than individuals from the same population. Yet sufficient genetic data can permit accurate classification of individuals into populations. To resolve this apparent conflict, we analyzed the question “How often is a pair of random individuals from two different populations genetically more similar than a pair of individuals randomly selected from any single population”; We compared this frequency (w) with error rates for classification methods, using data sets that vary in number of loci, diversity of populations, and polymorphism ascertainment strategies. Classification methods achieve higher discriminatory power than the individual-based measure, w, because of their use of aggregate properties of populations. The number of loci analyzed is the most critical variable: with one hundred polymorphisms, accurate classification is possible, but w remains sizable, even when using populations as distinct as sub-Saharan Africans and Europeans. Phenotypes controlled by a dozen or fewer loci can be expected to show substantial overlap between human populations. This provides empirical justification for caution when using population labels in biomedical settings, with broad implications for personalized medicine, pharmacogenetics, and the meaning of race.
This is the cautiously-worded Abstract to a Universtity of Utah paper downloadable in full here.
And this is what JW takes from it:-
This is crucially important, especially since it validates, essentially, EGI, while it is possible that a cursory examination of the paper by the “usual suspects” would lead them to an opposite, totally erroneous conclusion.
Reading the paper to its conclusion, the major finding is this: the idea that individuals from different (highly) distinct races could be more genetically similar than they would be to members of their own group is an illusion due to insufficient numbers of markers.
When 1000+ markers are used, Europeans, East Asians, and sub-Saharan Africans always are more genetically similar to members of their own group than to those of the other groups; the overlap in genetic similarity between these groups is ~ 0%.
It is true that the authors found, when “admixed” and “intermediate” groups are included, that the overlap does not quite reach zero, even with up to 10,000 markers, although the overlap does become very small.
However, three points (in increasing order of importance):-
1.It is possible that with even more markers, that the overlap between this more “continuous” sample of groups will reach ~ 0%, or at least become significantly lower than even the already very low overlap observed with 10,000 markers
2. EGI, by definition, is based upon ethnic-sized differences and concerns, dealing predominantly with large numbers of people. Even if there are outlier cases of overlap between certain groups, a member of any of these groups will always be more similar to their overall group than to the other overall group; this similarity, multiplied by population size will constitute the large differences in genetic interest described by Salter. Furthermore, even at the individual level, the overwhelming probability will be of intra-group greater similarity, so that a probalistic approach at the individual level will be roughly similar to the group level.
3. Most important, the EGI concept is based on relative differences in distinctive gene frequencies, not on overall genetic similarity. Certain gene variants may not show differences in frequency between groups, and these variants may contribute to the overlap in genetic similarity described by the authors. However, such variants do not contribute to differences in genetic interests between the groups, since the frequencies of the variants would not change, regardless of the outcome of competition between the groups. Even when considering a more “continuous” sampling of groups, it is highly unlikely that members of different groups would be more similar with respect to population distinctive gene frequencies than to members of their own groups, given a sufficient number of markers.
In summary, this paper strongly supports the EGI concept, particularly since the EGI concept is more stringent than a mere pairwise comparison of the similarity between individuals.
And, even such a mere pairwise comparison demonstrates the validity of the race concept; the idea promoted by certain ‘anti-racist’ activists that a white may be more similar to a black than to another white is proven incorrect. As long as sufficient markers are used, genetic similarity correlates very well with racial groupings.
In conclusion and perhaps for wider usage in future debate, these are the highlights, the money quotes, from the abstract and paper:-
From the Abstract
# The number of loci analyzed is the most critical variable.
From the Paper
# The power of large numbers of common polymorphisms is most apparent in the Microarray data set, comparing the European, East Asian and sub-Saharan African population groups (Figure 2C). ωˆ approaches zero (median 0.12%) with 1,000 polymorphisms. This implies that, when enough loci are considered, individuals from these population groups will always be genetically most similar to members of their own group
# The effect of population sampling becomes more pronounced when 1,000 or more loci are available. In the Microarray data set, ωˆ drops to zero at 1,000 loci if only distinct populations are sampled. With geographically intermediate and admixed populations added, however, ωˆ reaches an asymptotic value of 3.1%
# Even the pairwise relatedness measure, ωˆ, can show clear distinctions between populations if enough polymorphic loci are used
# Thus the answer to the question “How often is a pair of individuals from one population genetically more dissimilar than two individuals chosen from two different populations?”
# However, if genetic similarity is measured over many thousands of loci, the answer becomes “never” when individuals are sampled from geographically separated populations.
# Thus it may be possible to infer something about an individual’s phenotype from knowledge of his or her ancestry.
Comments (11) | Tell-a-Friend
The trackback URL for this entry is: Trackbacks are disabled for this entry
Trackbacks:
Comments:
Page 1 of 1 pages
So, the “hard work” to which I referred in
Removing Lewontin’s Fallacy from Hamilton’s Rule has begun.
Its great to see this happening.
How strange though that this must come from some podunk university among a bunch of inbred Mormons rather than the Ivy League!
My My My… it’s just BAFFLING!
PS: If you search for my article by that name via google, you won’t find the above linked article on MR.
That’s true. It comes up under “majorityrights.com Lewontin’s Fallacy Hamilton’s Rule”. But as soon as you replace the blog name with your own it’s gone.
Posted by Guessedworker on Sunday, March 18, 2007 at 11:56 PM | #
~~~~~Off-topic coment. ~~~~~
The following links to Lord Rothschild’s thinktank for Jewish Policy… click on “About” to read their agenda toward host societies.
http://www.jpr.org.uk/about/index.php
Posted by Marl Kalone on Monday, March 19, 2007 at 12:52 AM | #
The question that jumps into my mind is how do subgroups within the same race hold up under this analysis? They establish that Far Eastern and European peoples overlap by 0% if enough markers are used, but how much overlap is there between European and Near Eastern peoples?
Posted by President Barbicane on Monday, March 19, 2007 at 05:48 AM | #
Barbicane, if you read the post (and paper), you’ll see that the “w” for more closely related groups would be at least 3% - for the number of markers used in this paper.
Using more markers would drive that lower, and a whole genome analysis would likely drive it close to 0.
But, as the post states, “w” ignores structure, and it is not as stringent as EGI, which focuses on distinctive gene frequencies, as those that change with the outcome of genetic competition.
And, it is a numbers game, even if you use “w” and only the author’s markers. Given entire ethnic groups, even with low level overlap, you are heavily likely to be more similar to co-ethnics.
Posted by gnxp stinks on Monday, March 19, 2007 at 10:05 AM | #
From reading the “abstract” and “discussion” of this paper, it seems that the authors reveal their personal opinions on the race question. I note that they seem unhappy with their results and, rather than stress the conclusion that sufficient markers can distinguish individuals from distinct racial groups, they stress the illusion of overlap that exists if insufficient markers are measured. I’m no mind reader, but if I had to guess, I’d say that they are racial liberals, who wish that overlap would have been found between whites and blacks even at 10,000 markers and who are desperate to put forth whatever data to emphasize the illusory overlap. Is that why they chose to include the intermediate/admixed populations? Is that why they emphasize “rare” alleles with respect to phenotype, rather than mention that many phenotypic traits exhibit an Fst much higher than neutral genes (i.e., skin color)? One can almost imagine them begrudgingly, in a rage, writing the results for Europeans, Asians, and Africans with 1000+ markers.
In their “materials and methods” section, the authors write:
“Pairwise Genetic Distance
We use the “shared alleles” genetic distance (Bowcock et al. 1994; Chakraborty and Jin 1993; Mountain and Cavalli-Sforza 1997), which defines the distance between two individuals at a locus as one minus half the number of alleles they share. The genetic distance between individuals is the average of their per-locus distances. Pairs of individuals are classified as “within-population” or “between-population” according to whether the individuals were sampled from the same or different groups of populations as defined above.”
My understanding then is that they are looking at this from a “locus-by-locus” perspective (similar to Salter/Harpending) and averaging the results from the loci, each considered independently in turn. This would mean that they are NOT looking at the correlation structure from co-inherited alleles, which has been previously discussed on this blog as an important facet of human genetic variation and, hence, genetic interests.
I note also that while the authors do consider insertions, they do not consider copy number variation, inversions, or deletions, all of which can differ between population groups.
Thus, by ignoring “structural” considerations, they underestimate the differences between groups. Nevertheless, their data still demonstrate that the major racial groupings exhibit zero genetic similarity overlap of individuals from those groups, as long as sufficient markers are studied. And, even when the intermediate/admixed groups are added, and even in the absence of “structural” considerations, overlap is very low (and, possibly, even lower if more markers were used).
Posted by gnxp stinks on Monday, March 19, 2007 at 12:03 PM | #
Another point is that the only blog I have found (there may be others of course) that posted on this crucially important study before MR was:
http://yannklimentidis.blogspot.com/2007/03/genetic-similarities-within-and-between.html
I guess data that supports Salterism and the reality of race is not some people’s “cup of tea.”
Better to post pro-heterosis articles about white ranchers who happen to have the same NRY as some Negroes, and who foam at the mouth about the advantages on inter-breeding cattle.
Let’s get a “w” measurement of the rancher and his new-found Negro “family”; one may observe a lack of overlap when sampling from the autosomal genome.
The great Sailer tells us that Nick Wade of the NY Times is a stand-up guy on the reality of race; will Mr. Wade write on the FULL implications of the article.
Also, as I stated, we shouldn’t let the authors of this article “off the hook” just because they present solid data. Their “abstract” (especially) and their “discussion” are highly misleading.
Imagine someone lazy who just reads the abstract. They’ll come away with “...but w remains sizable, even when using populations as distinct as sub-Saharan Africans and Europeans. Phenotypes controlled by a dozen or fewer loci can be expected to show substantial overlap between human populations. “
Why not put the data from 1000+ markers and ~0% overlap and the words “the answer becomes “never” in the abstract? Why all the PC in the “discussion” including casting doubt on “race” when their data strongly supports the race concept?
They seem as bad as certain commercial gene companies, who offer products that emphasize the reality of race while, at the same time, writing against the race concept, etc.
Posted by gnxp stinks on Monday, March 19, 2007 at 12:55 PM | #
Klimentidis’ blog looks informative — just browsed it. I’m going to start checking it regularly.
Posted by Fred Scrooby on Monday, March 19, 2007 at 01:24 PM | #
The authors of ‘Genetic Similarities Within and Between Human Populations’ are clearly attempting to deal with the whole question in a genuinely scientific - as opposed to political- manner. Their conclusion that a pair of individuals from one population are “never” more dissimilar than two individuals taken from different populations when measured using a sufficiently large number of loci, proves this.
However, as has already been pointed out, the fact that this supports belief in an inherent & potentially meaningful difference between races is not mentioned; and indeed, is almost negated by stressing the greater similarity (as opposed to difference) between different populations, which is likely to be taken by the advocates of multiracialism as confirmation of their position.
In reality, to emphasize this is as misleading as pointing out the overwhelming ( 98%) similarity between human and chimpanzee DNA.
The other point to note is that while there is an acknowledgement of the different results which will be obtained when comparing “closely related” or “admixed” populations compared to “relatively (NB) isolated” ones, no details are given of exactly how much care has been taken to ensure that properly indigeneous peoples of different areas have been compared, as opposed to merely residents.
‘European’, etc, can have two distinct meanings, as can ‘American’ or ‘British’.
Posted by L Stewart on Tuesday, March 20, 2007 at 05:51 PM | #
If these guys are really toxic, I think what they may be attempting to do is something like a reductio ad absurdum to the genetic correlation structure argument along the following lines:
Other than clones, every person is genetically distinct. Hence, if we start with a known race for each person, it is trivial to construct a function that maps genotype to race with 100% accuracy. It is the lset of (genotype, race) ordered-pairs, indexed by genotype. Hence genetic correlation structure is meaningless.
This is known in machine learning as “over-training” or “over-fitting” and it basically amounts to not compressing—or not modeling—the data.
Of course, if you optimally compress your set of ordered pairs (your data) to its Komolgorov Complexity, you get an optimal model, similar to what is approximated by the cluster analysis software such as that used in:
http://www.journals.uchicago.edu/AJHG/journal/issues/v76n2/41839/41839.web.pdf
However you also get an exact model of how self-identified race relates to/differs from genetic correlation structure.
Posted by James Bowery on Tuesday, March 20, 2007 at 11:08 PM | #
Page 1 of 1 pages
Posted by James Bowery on Sunday, March 18, 2007 at 11:45 PM | #