Seeing the Structure for the SNPs

Posted by James Bowery on May 06, 2010, 10:10 AM | #

It really behooves our psychologists, such as MacDonald, to try to bring forth clinical treatments for the diseases that result in people knowing such things are true but feeling such truths are ‘bad’.  The psychological structure of such feelings is something we can blither about but really this is a very serious clinical psychology problem.  Jews certainly know how to create such mass psychopathology via mass media and academic promotion of men like Lewontin and they have dominated the 20th century’s psychotherapies leaving only their pathological definition of pathology, along with their pathogenic clinical treatments.

Posted by Dan Dare on May 06, 2010, 12:41 PM | #

Dasein - can you explain the difference between a haplotype and a haplotype cluster? Reason for asking is that it would be surprising to find that 12% of all haplotypes can be found in all regions/races. I should have thought that the only way in which that could be true is by taking recent trans-continental migration into account. By recent I mean since the beginning of European colonisation which has spread R1b worldwide.

Posted by Guessedworker on May 06, 2010, 05:05 PM | #

Dasein,

Am I right in thinking that this line of work could conceivably extrapolate and maybe even innumerate loss of genetic interest from loss of genetic structure, assuming that the preservation of distinct structure is itself an interest?

Posted by Dasein on May 06, 2010, 05:20 PM | #

Dan,

I’m not sure of the exact details of how the haplotype clusters were determined.  It was done using some software (fastPHASE).

From the Supplementary Materials:

Haplotypes and missing genotypes were estimated with fastPHASE 26 version 1.3, ordering SNPs on each chromosome according to positions from build 36.2 of the human genome sequence. As in Conrad et al.5, for estimating haplotypes and missing genotypes, geographic region labels (Table S5) were applied during the model fitting procedure to enhance accuracy (“population labels” in Scheet & Stephens 26). The number of haplotype clusters was set to 20, and we employed the default setting of 20 runs of the EM algorithm. This analysis was used to generate a “best guess” estimate of the true underlying patterns of haplotype structure.

And this is how it’s described in the paper:

For further assessment of haplotype variation, we devised a new
approach that avoided the difficulty of choosing window lengths for
haplotypic analysis. Variation is summarized locally at each point in
the genome by using a collection of 20 ‘haplotype clusters’, each of
which represents a group of haplotypes that overlap the point. For
every population, frequencies for the various haplotype clusters are
estimated at each SNP.

...

To make use of haplotypes in population structure analysis, we
generated ten haplotype cluster data sets, each of which assigned each
individual two haplotype clusters at every point along the genome,
with both cluster memberships ranging from 1 to 20. The ten data
sets were then analysed with the same methods as those used for
unphased genotypes, treating distinct clusters in the same manner
as distinct alleles.

I’m guessing these are like smaller haplogroups (smaller than R1b), and, like you suggest, if one were to somehow make these larger, they would become much more population-specific.  I don’t know exactly how fastPHASE decides what size a haplotype should be, I imagine it has to do with the correlation between SNPs, but I’d have to read the paper to say more (which I will probably do at some point, as it’s a topic in which I’ve become interested of late).

Posted by Dasein on May 06, 2010, 05:39 PM | #

GW,

That is the hope, even if this specific approach of using haplotypes (however defined) instead of individual SNPs isn’t the exact way it’s to be done.

Posted by PF on May 06, 2010, 08:50 PM | #

Dan wrote:

Dasein - can you explain the difference between a haplotype and a haplotype cluster? Reason for asking is that it would be surprising to find that 12% of all haplotypes can be found in all regions/races.

Haplotype is a vague and ad-hoc term for “combination of alleles at multiple loci that are transmitted together on the same chromosome” (wikipedia). Basically a string of SNPs of whatever length which are on the same chromosome.

Haplotype clusters look like they were determined in this paper by taking the genetic sequence data, and running algorithms which performed the following operation for each SNP in the 156 base pair sequence of the LCT. Beginning at the SNP at hand, the algorithm added neighboring SNPs into a string as it moved piecewise outwards, while simultaneously running comparisons on all emerging strings and forcing them to be resolved into 20 unique clusters using probabilistic (Bayesian?) analysis tools. (i.e. “I dont care if you’re two base pairs different from this other guy’s sequence, you’re probably related to him, given that you’re at least 10 base pairs different from everyone else’s sequences!”). The software cuts the ‘cluster’ (I presume) where it perceives sufficiently low linkage disequilibrium (nonrandom association) to qualify as a haplotype, which may be reflected in the probability calculations. Its a kernel smoother applied to genetic information.

It saves them from choosing a random stretch of DNA and saying ‘this is important, let us analyze the heritability of this here’, and thus circumventing the problem of ‘what qualifies as a haplotype?’ by providing an ‘any haplotype you like’ solution. 

Thanks for bringing in the hard science, Dasein.

Posted by James Bowery on May 06, 2010, 09:42 PM | #

PF writes:

The software cuts the ‘cluster’ (I presume) where it perceives sufficiently low linkage disequilibrium (nonrandom association) to qualify as a haplotype, which may be reflected in the probability calculations.

It sounds like they’re getting there.

Posted by PF on May 06, 2010, 11:17 PM | #

James wrote:

It sounds like they’re getting there.

Bioinformatics will find the units of selection you want, given a few more years. Proteomics will confirm the findings, and analysis like the kind which Dasein posted here will show us how do we differ in places where the differences matter. It wasn’t malignant interests (IME) that kept this analysis from happening, but the birth-pangs of bioinformatics approaches and the ghastly infinitude of our own devilish little sequences. Certainly malignant interests kept the wrong arguments in circulation forever, though.

Apparently our Jewish God played (another) trick on us. He composed the genetic code in the same way James Joyce composed Ulysses: 200 pages of garbage for every half-paragraph of insightful commentary. ACCT…TACT….TATAGCGCCG - get to the fucking point already!!

Posted by Praxis on May 07, 2010, 08:01 AM | #

Hs n Neanderthal Interbreeding:

http://www.washingtonpost.com/wp-dyn/content/article/2010/05/06/AR2010050604423_pf.html

Posted by Guessedworker on May 07, 2010, 08:16 AM | #

Perhaps Dasein will comment on the Neanderthal issue and its implications for genetic distinctiveness.

Posted by Søren Renner on May 07, 2010, 03:51 PM | #

http://johnhawks.net/weblog

Posted by James Bowery on May 08, 2010, 12:04 AM | #

PF writes: “It wasn’t malignant interests (IME) that kept this analysis from happening, but the birth-pangs of bioinformatics approaches and the ghastly infinitude of our own devilish little sequences…”

That depends on to what you are referring by “this analysis” of course.

“Apparently our Jewish God played (another) trick on us…. “

There really was no excuse for Boas hence Lewontin hence Gould at all.  Their ilk managed to set back sociobiology a century without any divine assistance.

Posted by Dasein on May 08, 2010, 06:18 AM | #

PF,

Part of the problem has indeed been the success of the ‘beanbag model’ in population genetics.  It may also be appropriate to speak of abortion, infanticide or child neglect that resulted in the dogmatic conflation of this model with descriptions of genetic diversity serving as an excuse for not quantifying difference at higher orders of genetic structure.  It has still as yet to be properly quantified (and consensus as to using such measures will surely meet plenty of resistance from politically motivated scientists), but at least the current data deluge has made it impossible to ignore these differences from a graphic point of view.  And James is right- there was nothing about the data that made this impossible to do many years ago.  Cavalli-Sforza got part way there with his pioneering use of PCA (which had some of its own problems), but the quantification of these differences remained fixed on single locus measures.  Despite the efforts of a few good doctors, academia has largely served as an abortion clinic for understanding of human diversity and genetic interests.

Posted by Dasein on May 08, 2010, 04:34 PM | #

GW,

I haven’t looked into the Neandertal paper in any detail.  Dienekes and John Hawks have written about it.  Like Dienekes, I find it puzzling that French, Han, and Papuans would have the same level of admixture.  Would sapiens mate with Neandertals in the Middle East, but then not when in close proximity over thousands of years in Europe to any detectable degree, as suggested by the authors?  Seems hard to believe.  In any case, however the admixture came about, I would be interested to know if there are differences in the Neandertal alleles which spread amongst Europeans vs. non-Europeans.  Because the environments were not the same (as well as demographic history), I would expect differences in the Neandertal alleles which spread in the populations (could be this is addressed in the paper). 

For racial nationalists, the good news is that Africans are now more different.  The system spin, though, will be that we’ve been mixing since the very early days (with different species!), so why the heck should anyone care about race-replacement?

Posted by Dasein on May 09, 2010, 07:29 AM | #

I should also point out that the “allelelic correlation structure” being considered here is of a relatively low level, namely that of stretches of adjacent SNPs along a single chromosome.  It is not considering the correlation between haplotype groups, whether on the same or different chromosomes.