"No scientific basis for race"

Note Added in response to 2020 Twitter mob attack which attempts to misrepresent my views:

This is not my research. The lead author affiliation for the paper discussed below is Harvard Medical School. I do not work on population structure or group differences in genomics.

This paper is from 2007. At the time even the capability to deduce ancestry ("race") from DNA was controversial. Now the technology is highly developed and used by millions of people (23andMe, Ancestry). The second part of the blog post discusses phenotype differences, not genetic causes of those differences. Genetic causes for phenotype differences could not yet be studied in 2008 and only now (circa 2020) is becoming an area of research that generates more light than heat.

Racist inferences based on the results of the paper are the fault of the reader, not the authors of the paper or of this blog.


##############################################


"It's just a social construction" -- a picture is worth a million words...




Caption: Each point is an individual, and the axes are two principal components in the space of genetic variation. Colors correspond to individuals of different European ancestry. (Via gnxp.)

The figure is from the following paper, reporting on a study of over 4000 individuals. The researchers can group most Europeans into a geographical cline (NW vs SE, that's the red band in the lower right of the figure; there are two clusters but also individuals who are in-between) + Ashkenazim (the pink isolated cluster in the upper left) using a few hundred markers. I'm sure even better resolution can be obtained with more loci.

Discerning the Ancestry of European Americans in Genetic Association Studies

Abstract: European Americans are often treated as a homogeneous group, but in fact form a structured population due to historical immigration of diverse source populations. Discerning the ancestry of European Americans genotyped in association studies is important in order to prevent false-positive or false-negative associations due to population stratification and to identify genetic variants whose contribution to disease risk differs across European ancestries. Here, we investigate empirical patterns of population structure in European Americans, analyzing 4,198 samples from four genome-wide association studies to show that components roughly corresponding to northwest European, southeast European, and Ashkenazi Jewish ancestry are the main sources of European American population structure. Building on this insight, we constructed a panel of 300 validated markers that are highly informative for distinguishing these ancestries. We demonstrate that this panel of markers can be used to correct for stratification in association studies that do not generate dense genotype data.

The money paragraph: "...Here we mine much larger datasets (more markers and more samples) to identify a panel of 300 highly ancestry-informative markers which accurately distinguish not just northwest and southeast European, but also Ashkenazi Jewish ancestry. This panel of markers is likely to be useful in targeted disease studies involving European Americans."

For previous discussion of genetic clustering of human populations, see here and here. It has been known for some time that major continental groups ("races") form distinct clusters. Improved data allow for much finer exploration of clusters within clusters.

This post is getting a lot of traffic from metafilter, and judging from the comments people are confused. I offer the following from the second link in the paragraph above:

...no matter what genetic markers you choose: SNPs, STRs, no matter how you choose them: randomly or based on their "informativeness", it is relatively easy to classify DNA into the correct continental origin. Depending on the marker types (e.g., indel vs. microsatellite), and their informativeness (roughly the distribution differences between populations), one may require more or less markers to achieve a high degree of accuracy. But, the conclusion is the same: after a certain number of markers, you always succeed in classifying individuals according to continental origin.

Thus, the emergent pattern of variation is not at all subjectively constructed: it does not deal specifically with visible traits (randomly chosen markers could influence any trait, or none at all), nor does it privilege markers exhibiting large population differences. The structuring of humanity into more or less disjoint groups is not a subjective choice: it emerges naturally from the genomic composition of humans, irrespective of how you study this composition. Rather than proving that race is skin-deep, non-existent, or unimportant, modern genetic science is both proving that it is in fact existent, but also sets the foundation for the study of its true importance, which is probably somewhere in between the indifference of the sociologists and the hyperbole of the racists.

One thing commenters seem particularly confused about is the difference between phenotypic and genetic variation. The clustering data show very clearly that, in certain subspaces, the genetic variation within a particular population cluster is less than between clusters. That is, the genetic "distance" between two individuals within a cluster is typically much less than the distance between clusters. (Technical comment: this depends on the number of loci or markers used. As the number gets large the distance between clusters becomes much larger than the individual cluster radius. For continental clusters, if hundreds or thousands of markers are used the intercluster distance dominates the intracluster size. Further technical comment: you may have read the misleading statistic, spread by the intellectually dishonest Lewontin, that 85% percent of all human genetic variation occurs within groups and only 15% between groups. The statistic is true, but what is often falsely claimed is that this breakup of variances (larger within group than between group) prevents any meaningful genetic classification of populations. This false conclusion neglects the correlations in the genetic data that are revealed in a cluster analysis. See here for a simple example which shows that there can be dramatic group differences in phenotypes even if every version of every gene is found in two groups -- as long as the frequency or probability distributions are distinct. Sadly, understanding this point requires just enough mathematical ability that it has eluded all but a small number of experts.) Update: see here for an explanation in pictures of Lewontin's fallacy. I also edited the paragraph above for clarity.

On the other hand, for most phenotypes (examples: height or IQ, which are both fairly heritable, except in cases of extreme environmental deprivation), there is significant overlap between different population distributions. That is, Swedes might be taller than Vietnamese on average, but the range of heights within each group is larger than the difference in the averages. Nevertheless, at the tails of the distribution one would find very large discrepancies: for example the percentage of the Swedish population that is over 2 meters tall (6"7) might be 5 or 10 times as large as the percentage of the Vietnamese population. If two groups differed by, say, 10 points in average IQ (2/3 of a standard deviation), the respective distributions would overlap quite a bit (more in-group than between-group variation), but the fraction of people with IQ above some threshold (e.g., >140) would be radically different. It has been claimed that 20% of all Americans with IQ > 140 are Jewish, even though Jews comprise only 3% of the total population.

...The imbalance continues to increase for still higher IQ’s. New York City’s public-school system used to administer a pencil-and-paper IQ test to its entire school population. In 1954, a psychologist used those test results to identify all 28 children in the New York public-school system with measured IQ’s of 170 or higher. Of those 28, 24 were Jews.

There is no strong evidence for specific gene variants (alleles) that lead to group differences (differences between clusters) in behavior or intelligence, but progress on the genomic side of this question will be rapid in coming years, as the price to sequence a genome is dropping at an exponential rate.

What seems to be true (from preliminary studies) is that the gene variants that were under strong selection (reached fixation) over the last 10k years are different in different clusters. That is, the way that modern people in each cluster differ, due to natural selection, from their own ancestors 10k years ago is not the same in each cluster -- we have been, at least at the genetic level, experiencing divergent evolution. In fact, recent research suggests that 7% or more of all our genes are mutant versions that replaced earlier variants through natural selection over the last tens of thousands of years.

European genetic substructure

Figure: Each point is an individual, and the axes are two principal components in the space of genetic variation. Colors correspond to individuals of different European ancestry.

The figure above is from the Nature paper: European Journal of Human Genetics (2008) 16, 1413–1429; doi:10.1038/ejhg.2008.210

Abstract: An investigation into fine-scale European population structure was carried out using high-density genetic variation on nearly 6000 individuals originating from across Europe. The individuals were collected as control samples and were genotyped with more than 300 000 SNPs in genome-wide association studies using the Illumina Infinium platform. A major East–West gradient from Russian (Moscow) samples to Spanish samples was identified as the first principal component (PC) of the genetic diversity. The second PC identified a North–South gradient from Norway and Sweden to Romania and Spain. ...

Some interesting points:

1) Significant East-West and North-South substructure is apparent already from the figure. The resolution of the study is sufficiently high that Swedes and Norwegians can be distinguished with 90 percent accuracy (Table 4). Crime scene forensics will never be the same -- "the Swede did it!" ;-)

In conclusion, we have shown that using PCA techniques it is possible to detect fine-level genetic variation in European samples. The genetic and geographic distances between samples are highly correlated, resulting in a striking concordance between the scatter plot of the first two components from a PCA of European samples and a geographic map of sample origins. We have shown how this information can be used to predict the origin of unknown samples in a rapid, precise and robust manner, and that this prediction can be performed without requiring access to the individual genotype data on the original samples of known origin. ...

2) Genetic distances between population clusters are roughly as follows: the distance between two neighboring western European populations is of order one in units of standard deviations and the distance to the Russian cluster is several times larger than that -- say, 3 or 4. From HapMap data, the distance from Russian to Chinese and Japanese clusters is about 18, and the distance of southern Europeans to the Nigerian cluster is about 19. The chance of mis-identifying a European as an African or E. Asian is exponentially small! (Table 5)

...The distance measure is a measure of the distance in standard deviations from a sample to the center of the closest matching population.

...For the other HapMap populations, the classification procedure assigned 100% of the YRI [Yoruban = Nigerian] samples to France, and almost 100% of the CHB and JPT [Chinese and Japanese] samples to Russia. However, the distribution of the distance measure for the four populations was quite different. For the CEU [HapMap European] samples, the median and 95% CI of the distance measure were 0.41 (0.11–1.01), whereas for the YRI, CHB and JPT populations, the median and 95% CIs were 19.3 (18.0–20.6), 17.7 (15.9–19.3) and 18.0 (15.4–19.6), respectively.

...The Yoruban [Nigerian] and Asian samples were identified as belonging to the countries on the south and east edges, respectively, of the European cluster, and the distance measure clearly indicates that they do not fit well into any of the proposed populations. ...

Figure: The three clusters shown above are European (top, green + red), Nigerian (light blue) and E. Asian (purple + blue).


See additional discussion at gnxp (the modified figure is from Razib), Dienekes

Related posts: "no scientific basis for race" , metric on the space of genomes

Learning can hurt

This essay, from 2007, summarizes how progress in genomics came to confront convenient but incorrect views on the genetic clustering of human populations. The vast majority of people, even scientists, are still confused about this subject. See also Human genetic variation, Fst and Lewontin's fallacy in pictures.

The flipside of serendipity: human genetics rediscovers race

... In this paper I investigate the recent re-emergence of genetic race in more detail, and endeavour to ascertain how, after a half-century hiatus, a young team of population geneticists could casually rediscover race in 2002. Was it happenstance? Or was it opportune timing? My tentative conclusions are: (a) the flipside of serendipity is sociopolitical context; and (b) in the first few years of the new millennium, on the face of it, the time was indeed ripe for genetic race to re-emerge within the scientific mainstream.

IT ALL STARTED WITH A CLUSTER ANALYSIS ...

'We have sequenced the genome of three females and two males, who have identified themselves as Hispanic, Asian, Caucasian or African American ... to help illustrate that the concept of race has no genetic or scientific basis. In the five Celera genomes, there is no way to tell one ethnicity from another.'--J Craig Venter, White House press conference, June 2000. (4)

Craig Venter was the 'maverick' geneticist who founded Celera Genomics--a private biotech company that, in the late 1990s, took on a trans-national network of publicly and privately funded researchers in a frantic race to sequence the human genome. In June 2000, at the White House press conference that heralded the completion of the 'first assembly' of the genome, standing next to his public-sector adversary Francis Collins, United States President Bill Clinton, and (via satellite feed) United Kingdom Prime Minister Tony Blair, in a triumphal moment for Big Science and liberal humanism Venter announced: 'the concept of race has no genetic or scientific basis.'

In saying this, he reaffirmed evolutionary biologist Richard Lewontin's oft-cited claim that racial classification is of 'virtually no genetic or taxonomic significance'. (5) He also echoed a gaggle of other geneticists and an army of anthropologists who, since the middle of the twentieth century, have similarly sought to deny that the 'race concept' has any basis in biology. In short, he took up what might be termed the 'race as social construct' position--a position that has come to function as a default setting within the social sciences in recent decades. Never mind that his assertion was based on an analysis of only five individual genomes and thus a statistical fallacy; the important thing was that it befitted the hyperbole of the occasion and had the aura of incontrovertible truth.

[[ Who could have possibly accepted these results as decisive? See Bounded Cognition. ]]

Over the next few years the incontrovertibility of the 'race as social construct' position began to erode. Just four years after Venter's speech, in the pages of a Nature genetics supplement dedicated to the issue of race in genomics, Francis Collins--who had shared both podium and Time magazine cover with Venter on that aforementioned momentous day--offered this cautious qualification: 'As ancestral origins in many cases have a correlation, albeit often imprecise, with self-identified race or ethnicity, it is not strictly true that race or ethnicity has no biological connection.' (6)

... Collins ... is the director of the US National Human Genome Research Institute, a renowned medical genetics researcher, and by all accounts not a racist. So why is a distinguished geneticist, not some racist crackpot on the margins of academia, suddenly making the claim: 'it is not strictly true that race or ethnicity has no biological connection'? What happened in those few short years that separate Venter's exultant assertion and Collins's wary disclaimer?

Ostensibly, the answer to these questions can be found in Collins's paper, for he cites evidence to support his claim: a population genetics study conducted by Noah Rosenberg and his team of researchers that was published in Science in December 2002. Drawing on samples from the Human Genome Diversity Cell Line Panel, (8) these researchers investigated the 'correspondence of predefined groups with those inferred from individual multilocus genotypes'. (9) They used a complex computer algorithm to sort 1,064 genome samples, from fifty-two different populations, on the basis of 4,199 different alleles, at 377 highly variable 'junk-DNA' loci, into varying numbers of statistically significant genetic clusters, and then compared the clusters with the geographical origins of the populations from which they were drawn. Put simply, they took the labels off the samples and tried to see if the computer could sort them back into meaningful groups based solely on their genetic similarities.

They found that 'predefined labels' (such as 'Yoruba', 'Italian' or 'Japanese') were 'highly informative about membership in genetic clusters'. (10) Further, when asked to identify five clusters, the computer grouped the samples into sets roughly corresponding to five geographical regions: (i) sub-Saharan Africa, (ii) Europe and West Asia, (iii) East Asia, (iv) Oceania, and (v) the Americas (see the row marked 'K=5' in Figure 1 below for a graphical representation of these clusters). Curiously, these regions are roughly geographically concordant to those occupied by the 'black', 'white', 'yellow', 'tawny' and 'copper-coloured' 'varieties' outlined in Johann Friedrich Blumenbach's seminal eighteenth-century racial typology. (11)

These results should have come as no surprise to most population geneticists, as it had long been assumed that human groups separated by physical, environmental, linguistic, and/or cultural barriers would display some degree of genetic differentiation. (12) Nonetheless, they were quite significant, as it was the first time that human populations had been comprehensively shown to cluster together on the basis of genetic likeness in a 'blind' test.

Still, Rosenberg and his colleagues were relatively circumspect in their conclusions. Their point was not to show that old anthropological conceptions of race are genetically 'real', but to argue that differentiating between human populations is both methodologically and statistically valid, and that such distinctions can be legitimately used for tracing the origins and migrations of peoples, as well as for medical and epidemiological purposes.

The words 'race' and 'ethnicity' never appeared in their paper, only 'population' and 'ancestry', and they were careful to choose neutral colours to code the continental clusters they identified--sub-Saharan Africa was orange, Europe and West Asia pale-blue, East Asia pink, Oceania green, and the Americas purple (not black, white, yellow, tawny and copper-coloured). Here they seemingly followed a convention established by the father of human population genetics and the prime mover behind the Human Genome Diversity Project, Luigi Luca Cavalli-Sforza, whose co-authored magnum opus, The History and Geography of Human Genes (1994, abridged 2006), featured a colour-coded atlas on its cover (Figure 2), with the caption: 'Four major ethnic regions are shown. Africans are yellow, Australians red, [Mongoloids blue] and Caucasoids green.' (13) A reviewer from Time magazine described Cavalli-Sforza's book as 'a landmark global study' that 'flattens The Bell Curve, proving that racial differences are only skin deep.' (14) Similarly, the Rosenberg study was lauded for its 'humanitarian' findings and adjudged 'paper of the year' by The Lancet in 2003. (15)

On the surface the consensus was that nothing much had changed. Population genetics had proven racial differences were only 'skin deep', and the Rosenberg findings only served to further confirm this incontrovertible fact. Everyone was content to repeat the familiar mantra: 'no race here ... nothing to look at ... move along'. However, beneath the 'race as social construct' ideological edifice a fuse had been lit and a 'race as biological reality' powder keg was about to go off.

THE MESSY FALLOUT

Even before the Rosenberg paper appeared, a group of belligerent geneticists headed up by Neil Risch from Stanford University had published an opinion piece in Genome Biology defending 'the validity of race/ethnicity categories for biomedical and genetic research' on the basis of cluster analyses similar to that used by the Rosenberg study.

[[ See Neil Risch, Portrait of a mathematical geneticist. ]]

The following year A W F Edwards, who had pioneered the statistical techniques behind cluster analysis with Cavalli-Sforza in the 1960s, penned a curt refutation of Lewontin's 1972 claim about the insignificance of racial classification, citing the Rosenberg study to buttress his argument. (17)

Following on from Edwards's assault, over the next few years several review essays appeared in major journals, (18) and the issue exploded onto the pages of the major broadsheets. Two articles from The New York Times tell the story of this re-emergence quite well: in August 2000 the headline read, 'Do races differ? Not really, genes show', with the author quoting Venter and others; (19) by July 2006 another headline read, 'Imperfect, imprecise but useful: Your race', and the article discussed the utility of racial classifications for biomedical research. (20)

A further twist in the plot of this recent race potboiler began on 14 March 2005, when evolutionary biologist Armand Marie Leroi published a now infamous op-ed piece, again in The New York Times (seemingly the arbiter of all things genetic), entitled 'A family tree in every gene'. (21) Taking up the Rosenberg study bait, he heralded their findings as suggesting: 'the consensus about social constructs was unravelling', and 'looked at the right way, the genetic data show that races clearly do exist.' (22) Further, he celebrated new advances in genetics that will soon mean, 'we shall no longer gawp ignorantly at the gallery [of racial differences]; we shall be able to name the painters.' (23)

Of course, more than a few people took exception to this talk of 'genetic painters' and 'racial galleries', and Leroi's article prompted a swift multi-pronged rebuttal. Within a month the US Social Science Research Council (SSCR) had sponsored a web forum collecting together critical responses from many of the usual 'social construct' suspects, including Alan Goodman, Evelynn Hammonds, Joseph Graves, Ruth Hubbard, Richard Lewontin and Jonathan Marks (many of whom also appeared in the 2003 PBS documentary series 'Race--the Power of an Illusion'). (24)

Labouring valiantly to plug the holes emerging in the epistemic dyke separating old race pseudoscience from modern human genetics, these ageing anti-race activists accused Leroi of 'reifying' race, and argued that race is most certainly not a biological category but rather an all-too-real social category that should be kept as far away as possible from the objective realm of science. ...


... This is what more sophisticated constructivists mean when they say race is a social construct--that the social and political aspects of race cannot easily be disentangled from the biological. (28) However, some constructivists want to go as far as to suggest that there is no biological dimension to race at all, with Richard Lewontin being a notorious culprit. Lewontin's research in the 1970s did not show that racial classifications were genetically insignificant, only that differences between racial groups were less significant than within groups. (29) His denial of the genetic significance of race is a non sequitur.

Putting aside Lewontin's fallacy, the major mistakes that unsophisticated social constructivists make are: (a) to suggest that being bound up with social and political concerns renders race 'unscientific' (as if scientific knowledge about human subjects could ever be separated from society and politics); and (b) to believe that by replacing race with the term population we can somehow escape this problem. The latter move just 'purifies' the concept of race, allowing geneticists to investigate racial groups such as 'Yoruba', 'Japanese', 'Caucasoid' or 'Australian' under the rubric of population, whilst denying that their research has anything to do with race. ...

PC censorship

I used to be an admirer of Berkeley economics professor Brad DeLong. I've been a regular reader of his blog for several years. Indeed, he, more than any other individual, inspired me to start blogging. This recent post on the US dollar conundrum shows him at his analytical and expository best.

I was shocked to discover that Professor DeLong actively and surreptitiously censors comments on his blog. You can read about it here. I can't explain his actions without assuming his goal is obfuscation, rather than truth seeking.

I've copied three of my comments below, originally posted to this discussion of human evolution. The third was removed without explanation.

Posted by: steve | Sep 12, 2005 7:55:47 PM

As mentioned, Brad's calculation neglects the possibility that mutations which might be adaptive in one environment are not necessarily adaptive in another.

According to the recent Lahn research (linked to by Anne, above), certain alleles of genes with direct effects on brain function have been subject to strong selection over the last tens of thousands of years. In certain populations (e.g., in Eurasia), the new alleles have rapidly replaced their predecessors, so they were clearly adaptive in those environments. In other populations (sub-Saharan Africa), the new alleles are rare, so they were probably less adaptive.

I don't know what will happen in the future, but current research shows that geographically isolated populations can have very different distributions of certain alleles, and not just those related to superficial features like skin or hair color.

Most fascinating is the possibility that relatively recent mutations had something to do with the rapid advancement of human civilization over the last 5000 years. (The ASPM variant may have emerged about that long ago.)

Posted by: steve | Sep 13, 2005 10:07:14 AM

Janet,

As already noted, it will take hundreds of years (at minimum) for mixing to eliminate the correlations between genes and "race" (or ancestral geographic lineage) that we currently have. That is a very long time from the perspective of social policy, although not in evolutionary terms.

I am by no means a fan of Sullivan, but I think he is correct to say that most liberals (I am one myself) have, due to wishful thinking, gratefully accepted the "there is no scientific basis for race" line. Anne's post of the NYTimes op-ed by LeRoi gives the history of this facile, but now doomed, position. (Cochran's explanation above is very clear - better than LeRoi's.) I don't think most people appreciate that we are now on a Moore's Law growth curve for genomic information. Google "hapmap" and have a look for yourself at the state of the art.

Rather than rely on the scientifically unsupported claim that "we are all equal," it would be better to teach our students that we all have inalienable human rights regardless of our abilities or genetic make up. Continuing to rely on the false equality premise only undermines the liberal position on race issues.

Posted by: steve | Sep 13, 2005 11:17:21 PM

gcochran wrote:

"Do principal component analysis on the covariance matrix for many loci (or cluster analysis) and !presto! - Bob's your uncle."

This gets right to the point (see an earlier post by gcochran for a less terse explanation). Too bad that very few readers here will understand (or even try to understand) what it means. Bambi vs Godzilla had the insight to ask the question properly. Will he or she make the effort to understand the answer?

Imagine each individual's genetic code as a point in a space of *very high* dimension. Then look at clusters of points. (Define a cluster as a group of points whose distance from each other is less than some radius; distinct clusters are separated by distances larger than this radius.) These clusters map directly onto traditional groupings of ethnicity. In fact, a recent study by Neil Risch at UCSF showed that self-reported "race" correlates very well with the clustering results. (Mixed race people are obviously an exception, but as discussed they are a small fraction of the total population, and will continue to be for some time.)

People (especially professors of social science) who confidently state to their students that "there is no genetic basis for race" should think through the analysis described above and look at the data carefully if they want to retain their credentials as scientists.

From the conclusions of the Risch paper (Am. J. Hum. Genet. 76:268–275, 2005):

Attention has recently focused on genetic structure in the human population. Some have argued that the amount of genetic variation within populations dwarfs the variation between populations, suggesting that discrete genetic categories are not useful (Lewontin 1972; Cooper et al. 2003; Haga and Venter 2003). On the other hand, several studies have shown that individuals tend to cluster genetically with others of the same ancestral geographic origins (Mountain and Cavalli-Sforza 1997; Stephens et al. 2001; Bamshad et al. 2003). Prior studies have generally been performed on a relatively small number of individuals and/or markers. A recent study (Rosenberg et al. 2002) examined 377 autosomal micro-satellite markers in 1,056 individuals from a global sample of 52 populations and found significant evidence of genetic clustering, largely along geographic (continental) lines. Consistent with prior studies, the major genetic clusters consisted of Europeans/West Asians (whites), sub-Saharan Africans, East Asians, Pacific Islanders, and Native Americans. ethnic groups living in the United States, with a discrepancy rate of only 0.14%.

Spacetime topology change and black hole information

I'm in Erice right now, so posting may (or may not) be interrupted depending on how exciting the talks are :-)

hep-th/0608175

Spacetime topology change and black hole information

Abstract: Topology change --- the creation of a disconnected baby universe --- due to black hole collapse may resolve the information loss paradox. Evolution from an early time Cauchy surface to a final surface which includes a slice of the disconnected region can be unitary and consistent with conventional quantum mechanics. We discuss the issue of cluster decomposition, showing that any violations thereof are likely to be unobservably small. Topology change is similar to the black hole remnant scenario and only requires assumptions about the behavior of quantum gravity in planckian regimes. It does not require non-locality or any modification of low-energy physics.


I don't necessarily want to claim that topology change is plausible. But, it does suggest that we can't make progress on the black hole information problem by low-energy means alone, since there are possible solutions that have unobservable low-energy effects and depend on how quantum gravity behaves at planck densities.

I got some email already pointing me to the following interesting papers on this issue. The first two are by Ted Jacobson and the last by Easson and Brandenberger. All discuss, in one way or another, the possibility that things which fall into black holes end up "somewhere else" and the relation to the information loss paradox.

gr-qc/0308048
gr-qc/9908031
hep-th/0103019

More. Early work by Mukhanov and Frolov on connecting deSitter space to the black hole interior (they assume strong planckian dynamics leads to a symmetrical equation of state with stress tensor proportional to g_mn, so deSitter is natural): http://prola.aps.org/abstract/PRD/v41/i2/p383_1 (too old for arXiv).

There are also apparently many loop QG papers suggesting a repulsive equation of state at planck density, which is what is needed to have expansion instead of a singularity. I haven't tracked those papers down, but will add them here and to the revision of the paper. Also, Lee Smolin long ago proposed that black hole collapse might lead to new universe creation, although I believe that was part of a cosmological selection proposal and not to solve the information loss problem.

Interestingly, there seem to be two camps on this issue. People from a string theory background seem convinced that only the two traditional solutions are viable: locality violation and quantum xeroxing at the semiclassical horizon (possibly due to stringy effects and bh complementarity), or remnants. In the relativist community it seems people find the topology change option rather obvious and wonder why I wasn't aware of these earlier references. I'd like to put the two sides in a seminar room somewhere to mix it up! :-)

I came to this from the stringy side -- the review articles I cite are by Strominger, Banks, Giddings and Susskind, so I thought arguing the feasibility of the topology change scenario to be an uphill fight. The very first thing I do in the paper is present the two main objections to the idea. The first objection, that pure to mixed evolution (even at an effective level) must entail energy non-conservation, seems to have been discussed before by relativists (e.g., Wald). But the second objection, regarding cluster decomposition violation (multiple baby universe creation amplitudes don't factorize even when the black hole collapses are widely separated) does not seem to have been widely discussed.

John McWhorter: the truth about mismatch

I'm shocked that CNN published Columbia professor John McWhorter's editorial on Scalia's mismatch comments. His remarks challenge the mainstream media narrative, and require some thought from the reader.

CNN: Those who consider themselves on black people's side are having a field day dismissing Justice Antonin Scalia as a racist. His sin was suggesting that black students admitted to the most selective institutions might perform better at somewhat less selective institutions where instruction is paced more slowly.

I don't usually agree with Justice Scalia's perspectives, but we are doing him wrong on this one. Scalia didn't express himself as gracefully as he could have. No one could suppose that anything like all black students find the pedagogical pace at top-level universities overwhelming.

However, Scalia's comment stemmed not from random intuition but from research showing that a substantial number of black students would do better -- and be happier -- at schools less selective than the ones they are often admitted to via racial preferences.

The reading public's response to Scalia's point shows that few have any idea of this research or assume it was done by partisan zealots. An intelligent discussion of the Fisher v. University of Texas case now before the Supreme Court requires a quick tour of the facts.

... At Duke University, economist Peter Arcidiacono, with Esteban Aucejo and Joseph Hotz, has shown that the "mismatch" lowers the number of black scientists. Black students at a school where teaching is faster and assumes more background than they have often leave the major in frustration, but would be less likely to have done so at a school prepared to instruct them more carefully.

UCLA law professor Richard Sander conclusively showed in 2004 that "mismatched" law students are much more likely to cluster in the bottom of their classes and, especially, to fail the bar exam. Meanwhile, Sander and Stuart Taylor's book argues that the mismatch problem damages the performance of black and brown students in general.

There are scholars who dispute Sander and Taylor's thesis about undergraduate school in general. However, when it comes to the more specific points about STEM subjects and law school, takedown arguments are harder to fashion because of the simple force of the facts.

For example, on Sander's widely publicized law school paper, time has passed and few of us go in for reading law review articles. However, Emily Bazelon's widely read critique of it was hasty in claiming that the responses published along with Sander's piece refuted his claims. Rather, anyone reading them with an open mind would see that they left Sander's basic point standing tall and this applies to any other critique I have seen: there has been no "smackdown." It is similarly unlikely that anyone could tell Arcidiacono, Aucejo and Hotz that what they chronicle was mirages.

... At the University of California, San Diego the year before racial preferences were banned in the late '90s, exactly one black student out of 3,268 freshmen made honors. A few years later, after students who once would have been "mismatched" to flagship school UC Berkeley were now admitted to schools such as UC San Diego, one in five black freshmen were making honors, the same proportion as white ones.

What civil rights leader of the past would have seen this as racism? Who in the future will? Or why are we tarring Scalia as a bigot for espousing outcomes like this in the here and now?

Our national conversation on racial preferences is under-informed and mean when founded on an assumption that anyone who seriously questions racial preferences is naive at best and a pig at worst. Affirmative action is a complex matter upon which reasonable minds will differ. With the well-being of young people of color at stake, we can't afford to pretend otherwise.

You could also have read about this topic in my NYTimes op-ed from 2012: Merit, Not Race, in College Admissions. The facts supporting mismatch are not disputable, despite the attempts of some ideologues to cloud the conversation.

I sometimes explain the issue as follows. Imagine taking a group of typical engineering students from Iowa State University and transferring them to MIT or Caltech in their freshman year. What are the odds that these students would thrive? What are the odds that they would cluster at the bottom of the class and learn less than they would have had they stayed at Iowa State? Anyone who has taught STEM at both highly selective and less selective universities knows that large differences in admissions selectivity lead to large differences in average ability in the classroom (the whole purpose of selection!), and that the pace and presentation of material needs to be adjusted accordingly. In the example I gave, the SAT gap is perhaps 200 points (on a 1600 scale). But this is smaller than the admissions preference given to African Americans by most selective colleges (see Affirmative Action: the Numbers).

Why do we think the thought experiment would suddenly become a good idea if the students were black?

One might object that SAT or high school GPA are flawed measures of ability, especially for under-represented minorities. But this has been studied carefully. The fact of the matter is that the accuracy of these numerical indicators as predictors of college performance varies little depending on the race or even socio-economic background of the student. (To be technical: adding an additional variable for race or income to the regression changes the SAT coefficient by very little.) That is, a student admitted with lower scores than their peers is unlikely to perform well in difficult STEM majors, regardless of the race of the student, and even if that student comes from a wealthy legacy family.

It's also known that test preparation only improves SAT scores by a small fraction of the typical admissions preference (i.e., less than 50 points vs 300), and that test-retest reliability of the exam is very high. The tests measure something real, which has predictive power.

The consequences of selecting students based on academic ability are clearly manifested in this study by Jonathan Wai and myself: Colleges ranked by Nobel, Fields, Turing and National Academies output. Colleges with the most talented students (selected based on simple measures such as SAT and high school GPA) produce orders of magnitude more top scientists, engineers, and medical researchers per capita than less selective schools.

There is a sad pattern in the comprehensive Duke data that both McWhorter and I reference: students admitted with weak admissions scores are more likely to leave challenging STEM majors in favor of less competitive subjects, and they are more likely to perform poorly overall. This pattern holds regardless of the race or socio-economic status of the student.

Anyone who claims to have a serious interest in higher education (e.g., all professors and administrators) should be familiar with the facts presented above.

Note Added:

Two related editorials, one by Richard Sander in the WSJ and the other by Thomas Sowell.

Most of the analyses attacking mismatch have focused on graduation rates. But these ignore the fact that virtually all colleges have easy majors. Given the widely acknowledged practice of admitting wealthy applicants, legacies, and athletes with significantly below average scores, and the nearly 100% graduation rate at the Ivies, the conclusion has to be that there are paths of little resistance through most elite colleges. Surprisingly, it might be easier for a student of average ability (that is, relative to the overall population) to graduate from Harvard (once admitted), than to graduate from a typical state university -- the key is choice of major.

IIRC, the migration of weaker students from STEM into less challenging majors is revealed in "real time" in the Duke data set, which contains every grade in every course for all students over a number of years. It also contains the application files of students and their original intended majors. Without this level of specificity, one can't really make strong assertions about mismatch. I have yet to see any data-driven analyses which contradict the mismatch hypothesis specifically as it applies to STEM or law school.

We should applaud Historically Black Colleges and Universities (HBCUs) for taking students with modest test scores and yet producing graduates who go on to earn a disproportionate fraction of STEM PhDs, something Scalia alluded to in his (badly worded) remarks.

NSF: ... Among known U.S. baccalaureate-origin institutions of 1997–2006 black S&E doctorate recipients, the top 8 and 20 of the top 50 were HBCUs. ... The top 5 baccalaureate-origin institutions of 1997–2006 black S&E doctorate recipients were: Howard University, Spelman College, Hampton University, Florida A&M University, and Morehouse College.

Metric on the space of genomes and the scientific basis for race

[Note: the science has advanced considerably since I wrote this post. See here for a nice picture of clustering by ethnicity. I've added some comments at the bottom.]

Suppose that the human genome has 30,000 distinct genes, which we will label as i = 1,2, ... N, where N = 30k. Next, suppose that there are n_i variants or alleles (mutations) of the i-th gene. Then, each human's genetic information can be described as a point on a lattice of size n_1 x n_2 x n_3 ... n_N, or equivalently an N-tuple of integers, each of whose values range from 1 to n_i. For the simplified case where there are exactly 10 variants of each gene, the number of points in this N dimensional space is 10^N or 10^{30k}, one for each distinct 30k digit number. It's a space of very high dimension, but this doesn't stop us from defining a metric, or measure of distance between any two points in the space. (For simplicity we ignore restrictions on this space which might result from incompatibility of certain combinations, etc.)

Note that the genomes of all of the humans who have ever lived occupy only a small subset of this space -- most possible variations have never been realized. For this reason, the surprise expressed by biologists that humans have so few genes (not many more than a worm, and far less than the 100k of earlier estimates) is no cause for concern -- the number of possible organisms that might result from 30k genes is enormous -- far more than the number of molecules in the visible universe.

To define a metric, we need a notion of how far apart two different alleles are. We can do this by counting base pair differences -- most mutations only alter a few base pairs in the genetic code. We can define the distance between two alleles in terms of the number of base pair changes between them (this is always a positive number). Then, we can define the distance between two genomes as the sum of each of the i=1,2,..,N individual gene distances. It is natural, although perhaps not always possible, to choose the n_i labeling of alleles to reflect relative distances, so variants n_1 and n_2 are close together, and both very far from n_10.

The exact definition of the metric and the allele labeling are somewhat arbitrary, but you can see it is easy to define a meaningful measure of how far apart any two individuals are in genome space.

Now plot the genome of each human as a point on our lattice. Not surprisingly, there are readily identifiable clusters of points, corresponding to traditional continental ethnic groups: Europeans, Africans, Asians, Native Americans, etc. (See, for example, Risch et al., Am. J. Hum. Genet. 76:268–275, 2005.) Of course, we can get into endless arguments about how we define European or Asian, and of course there is substructure within the clusters, but it is rather obvious that there are identifiable groupings, and as the Risch study shows, they correspond very well to self-identified notions of race.

From the conclusions of the Risch paper (Am. J. Hum. Genet. 76:268–275, 2005):

Attention has recently focused on genetic structure in the human population. Some have argued that the amount of genetic variation within populations dwarfs the variation between populations, suggesting that discrete genetic categories are not useful (Lewontin 1972; Cooper et al. 2003; Haga and Venter 2003). On the other hand, several studies have shown that individuals tend to cluster genetically with others of the same ancestral geographic origins (Mountain and Cavalli-Sforza 1997; Stephens et al. 2001; Bamshad et al. 2003). Prior studies have generally been performed on a relatively small number of individuals and/or markers. A recent study (Rosenberg et al. 2002) examined 377 autosomal micro-satellite markers in 1,056 individuals from a global sample of 52 populations and found significant evidence of genetic clustering, largely along geographic (continental) lines. Consistent with prior studies, the major genetic clusters consisted of Europeans/West Asians (whites), sub-Saharan Africans, East Asians, Pacific Islanders, and Native Americans. ... We have shown a nearly perfect correspondence between genetic cluster and SIRE [self-reported ethnicity] for major ethnic groups living in the United States, with a discrepancy rate of only 0.14%.

This clustering is a natural consequence of geographical isolation, inheritance and natural selection operating over the last 50k years since humans left Africa.

Every allele probably occurs in each ethnic group, but with varying frequency. Suppose that for a particular gene there are 3 common variants (v1, v2, v3) all the rest being very rare. Then, for example, one might find that in ethnic group A the distribution is v1 75%, v2 15%, v3 10%, while for ethnic group B the distribution is v1 2% v2 6% v3 92%. Suppose this pattern is repeated for several genes, with the common variants in population A being rare in population B, and vice versa. Then, one might find a very dramatic difference in expressed phenotype between the two populations. For example, if skin color is determined by (say) 10 genes, and those genes have the distribution pattern given above, nearly all of population A might be fair skinned while all of population B is dark, even though there is complete overlap in the set of common alleles. Perhaps having the third type of variant v3 in 7 out of 10 pigmentation genes makes you dark. This is highly likely for an individual in population B with the given probabilities, but highly unlikely in population A.

We see that there can be dramatic group differences in phenotypes even if there is complete allele overlap between two groups - as long as the frequency or probability distributions are distinct. But it is these distributions that are measured by the metric we defined earlier. Two groups that form distinct clusters are likely to exhibit different frequency distributions over various genes, leading to group differences.

This leads us to two very different possibilities in human genetic variation:

Hypothesis 1: (the PC mantra) The only group differences that exist between the clusters (races) are innocuous and superficial, for example related to skin color, hair color, body type, etc.

Hypothesis 2: (the dangerous one) Group differences exist which might affect important (let us say, deep rather than superficial) and measurable characteristics, such as cognitive abilities, personality, athletic prowess, etc.

Note H1 is under constant revision, as new genetically driven group differences (e.g., particularly in disease resistance) are being discovered. According to the mantra of H1 these must all (by definition) be superficial differences.

A standard argument against H2 is that the 50k years during which groups have been separated is not long enough for differential natural selection to cause any group differences in deep characteristics. I find this argument quite naive, given what we know about animal breeding and how evolution has affected the (ever expanding list of) "superficial" characteristics. Many genes are now suspected of having been subject to strong selection over timescales of order 5k years or less. For further discussion of H2 by Steve Pinker, see here.

The predominant view among social scientists is that H1 is obviously correct and H2 obviously false. However, this is mainly wishful thinking. Official statements by the American Sociological Association and the American Anthropological Association even endorse the view that race is not a valid biological concept, which is clearly incorrect.

As scientists, we don't know whether H1 or H2 is correct, but given the revolution in biotechnology, we will eventually. Let me reiterate, before someone labels me a racist: we don't know with high confidence whether H1 or H2 is correct.

Finally, it is important to note that group differences are statistical in nature and do not imply anything definitive about a particular individual. Rather than rely on the scientifically unsupported claim that we are all equal, it would be better to emphasize that we all have inalienable human rights regardless of our abilities or genetic makeup.

[See here (Economist's View blog) for more comments.]

[See Gene Expression for more discussion and references.]

Note added: See related Risch article which discusses H1 and H2: Assessing genetic contributions to phenotypic differences among 'racial' and 'ethnic' groups (Nature 2004), and also these 2013 comments in an interview with Ta-Nehisi Coates for The Atlantic:

Q: One last question. Your paper on assessing genetic contributions to phenotype, seemed skeptical that we would ever tease out a group-wide genetic component when looking at things like cognitive skills or personality disposition. Am I reading that right? Are "intelligence" and "disposition" just too complicated?

A: (Risch) Joanna Mountain and I tried to explain this in our Nature Genetics paper on group differences. It is very challenging to assign causes to group differences. As far as genetics goes, if you have identified a particular gene which clearly influences a trait, and the frequency of that gene differs between populations, that would be pretty good evidence. But traits like "intelligence" or other behaviors (at least in the normal range), to the extent they are genetic, are "polygenic." That means no single genes have large effects -- there are many genes involved, each with a very small effect. Such gene effects are difficult if not impossible to find. The problem in assessing group differences is the confounding between genetic and social/cultural factors. If you had individuals who are genetically one thing but socially another, you might be able to tease it apart, but that is generally not the case.

In our paper, we tried to show that a trait can appear to have high "genetic heritability" in any particular population, but the explanation for a group difference for that trait could be either entirely genetic or entirely environmental or some combination in between.

So, in my view, at this point, any comment about the etiology of group differences, for "intelligence" or anything else, in the absence of specific identified genes (or environmental factors, for that matter), is speculation.

To repeat, I think it is fair to say that we don't know*** whether H1 or H2 is correct, even though we do know that populations cluster (genetically) by ancestry. Clustering makes it possible that H2 is correct, because the alleles (genetic variants) affecting a particular phenotype will tend to have different frequencies in different groups. The average value of the trait might or might not be the same in different groups. In the case of height, enough is known to suggest that variants which lead to increased height are more frequent in northern Europe than in the south, and there is evidence that this is due to selection, not drift.

*** I think it's especially important to be epistemologically careful in thinking about these matters, because of our difficult history with race. I would much rather live in a world where H1 is true and H2 false. But my preference alone does not make it so. (I would also much rather live in a universe created by a loving God, and in which I and my children have eternal souls; not a cruel Darwinian universe in which our species arose merely by chance. But my preference does not make it so.)

SubAltern Homesick Blues

New York Magazine seems to have dedicated an entire issue to the Alt-Right. If you don't recognize any of this subterranean internet stuff you should probably have a look. Somehow they left out the Illuminati, though.

See The Paranoid Style in American Politics and Subaltern Postcolonial Gramsci Homi Bhabha Babble.

BEYOND ALT: THE EXTREMELY REACTIONARY, BURN-IT-DOWN-RADICAL, NEWFANGLED FAR RIGHT

... what follows here is an attempt to really reckon with the alt-right and its fellow travelers: to organize and catalogue influences, philosopher-kings, and shit-posting foot soldiers; to track the movement’s history, its future, and the story of how the modern internet made it possible; to study its grievances, its media savvy, its symbols, its heroes and villains, its president and its critics of the president, its billionaire supporters and the underemployed message-board-dwelling “advocates” who serve as its creative engine. The movement is not a monolith — though it would also never be mistaken for a rainbow coalition — and part of what we’ve focused on is just how the various wings work together in concert. How does Steve Bannon relate to Russia Today, and what do conspiracy theorists have in common with pickup artists and Nazi Furries? How do memes like milk become weaponized, and just when did 4chan get political? How do its beliefs get amplified into the larger culture?

For all its theoretical anti-modernity, the alt-right is a uniquely modern movement, after all, enabled in reach by the connective tools of the social internet and by the clever use of the sort of irony that every young American raised on TV and memes recognizes as our pop-culture lingua franca. It’s an irony they’ve used as armor, too: If you take them seriously, they’ll claim you missed the joke. But of course, by treating them as a joke, you can miss their importance ...

Spooooky!

The Techno-Libertarians Praying for Dystopia

... people who believe that the future of our species involves shedding our humanity in a marriage with AI; this is known as transhumanism, and it has not unreasonably been called a new tech religion. Though the movement has no explicit political affiliations, it tends, for reasons that are probably self-explanatory, to draw a disproportionate number of Silicon Valley libertarians. And the cluster of ideas at its center — that the progress of technology will inevitably render good ol’ Homo sapiens obsolete; that intelligence, pure computational power, is to be pursued above all other values — has exerted a powerful attraction on a small group of futurists whose extreme investment in techno-libertarianism has pushed them over an event horizon into a form of right-wing authoritarianism it might be useful to regard as Dark Transhumanism.

PLINK 1.90 alpha

WDIST is now PLINK 1.9 alpha. WDIST (= "weighted distance" calculator) was originally written to compute pairwise genomic distances. The mighty Chris Chang then amazingly re-implemented all of PLINK with significant improvements (see below).

PLINK 1.9 even has support for LASSO (i.e., L1 penalized optimization, a particular method for Compressed Sensing).

This is a comprehensive update to Shaun Purcell's popular PLINK command-line program, developed by Christopher Chang with support from the NIH-NIDDK's Laboratory of Biological Modeling and others. (What's new?) (Credits.)

It isn't finished yet (hence the 'alpha' designation), but it's getting there. We are working with Dr. Purcell to launch a large-scale beta test in the near future. ...

Unprecedented speed 

Thanks to heavy use of bitwise operators, sequential memory access patterns, multithreading, and higher-level algorithmic improvements, PLINK 1.9 is much, much faster than PLINK 1.07 and other popular software. Several of the most demanding jobs, including identity-by-state matrix computation, distance-based clustering, LD-based pruning, and association analysis max(T) permutation tests, now complete hundreds or even thousands of times as quickly, and even the most trivial operations tend to be 5-10x faster due to I/O improvements.

We hasten to add that the vast majority of ideas contributing to PLINK 1.9's performance were developed elsewhere; in several cases, we have simply ported little-known but outstanding implementations without significant further revision (even while possibly uglifying them beyond recognition; sorry about that, Roman...). See the credits page for a partial list of people to thank. On a related note, if you are aware of an implementation of a PLINK command which is substantially better what we currently do, let us know; we'll be happy to switch to their algorithm and give them credit in our documentation and papers.

Nearly unlimited scale 

The main genomic data matrix no longer has to fit in RAM, so bleeding-edge datasets containing tens of thousands of individuals with exome- or whole-genome sequence calls at millions of sites can be processed on ordinary desktops (and this processing will usually complete in a reasonable amount of time). In addition, several key individual x individual and variant x variant matrix computations (including the GRM mentioned below) can be cleanly split across computing clusters (or serially handled in manageable chunks by a single computer).

Command-line interface improvements
We've standardized how the command-line parser works, migrated from the original 'everything is a flag' design toward a more organized flags + modifiers approach (while retaining backwards compatibility), and added a thorough command-line help facility.

Additional functions
In 2009, GCTA didn't exist. Today, there is an important and growing ecosystem of tools supporting the use of genetic relationship matrices in mixed model association analysis and other calculations; our contributions are a fast, multithreaded, memory-efficient --make-grm-gz/--make-grm-bin implementation which runs on OS X and Windows as well as Linux, and a closer-to-optimal --rel-cutoff pruner.

There are other additions here and there, such as cluster-based filters which might make a few population geneticists' lives easier, and a coordinate-descent LASSO. New functions are not a top priority for now (reaching 95%+ backward compatibility, and supporting dosage/phased/triallelic data, are more important...), but we're willing to take time off from just working on the program core if you ask nicely.

Human genetic variation, Fst and Lewontin's fallacy in pictures

In an earlier post European genetic substructure, I displayed the following graphic, illustrating the genetic clustering of human populations.




Figure: The three clusters shown above are European (top, green + red), Nigerian (light blue) and E. Asian (purple + blue).

The figure seems to contradict an often stated observation about human genetic diversity, which has become known among experts as Lewontin's fallacy: genetic variation between two random individuals in a given population accounts for 80% or more of the total variation within the entire human population. Therefore, according to the fallacy, any classification of humans into groups ("races") based on genetic information is impossible. ("More variation within groups than between groups.")

To understand this statement better, consider the F statistic of population genetics, introduced by Sewall Wright:

Fst = 1 - Dw / Db

Db and Dw represent the average number of pairwise differences between two individuals sampled from different populations (Db = "difference between") or the same population (Dw = "difference within"). Even in the most widely separated human populations Fst < .2 so Dw / Db > .8 (roughly). This may not sound like very much genetic diversity, but it is more than in many other animal species. See here for recent high statistics Fst values by nationality.

Dw / Db > .8 means that the average genetic distance measured in number of base pair differences between two members of a group (e.g., two randomly selected Europeans) is at least 80 percent of the average distance between distant groups (e.g., Europeans and Asians or Africans). In other words, if two individuals from very distant groups (e.g., a Japanese and a Nigerian) have on average N base pair differences, then two from the same group (e.g., two Nigerians or two Japanese) will on average have roughly .8 N base pair differences.

How can the Fst result ("more variation within groups than between groups") be consistent with the clusters shown in the figure? I've had to explain this on numerous occasions, always with great difficulty because the explanation requires a little mathematics. In order to make the point more accessible, I've created the figures below, which show two population clusters, each represented by an ellipsoid (blob). The different figures depict the same pair of objects, just viewed from different angles.

The blobs are constructed and arranged so that the average distance between two points (individuals) within the same cluster is almost as big as the average distance between two points (individuals) in different clusters. This is easy to achieve if the ellipsoids are big and flat (like pancakes) and placed close to each other along the flat directions. The figure is meant to show how one can have small Fst, as in humans, yet easily resolved clusters. The direction in which the gap between the clusters appears is one of the principal components in the space of human genetic variation, as recently found by bioinformaticists. The figure at the top of this post plots individuals as points in the space generated by the two largest principal components extracted from the combination of data from HapMap and from large statistics sampling of Europeans. Exhibited this way, isolated clusters ("races") are readily apparent.

The real space of genetic variation has many more than 3 dimensions, so it can't be easily visualized. But some aspects of the figures below still apply: there will be particular directions of variation over which different populations are more or less identical (orthogonal to the principal component; i.e. along the flat directions of each pancake), and there will be directions in which different populations differ radically and have little or no overlap. Note, however, that we are specifically referring to genetic variation, which may or may not translate into phenotypic variation.







Related posts: "no scientific basis for race" , metric on the space of genomes.

The existence of this clustering has been known for 40 years.

Power laws, baby

Thanks to reader GS for the link to this excellent article by statistical physicist Eugene Stanley. There's much more to the article, but I decided to excerpt this discussion of security price fluctuations and the observation that they are far from log normal.

Stanley: ...

When they analyzed these data–200 million of them–in exactly the same fashion that Bachelier had analyzed data almost a century earlier, they made a startling discovery. The pdf of price changes was not Gaussian plus outliers, as previously believed. Rather, all the data–including data previously termed outliers–conformed to a single pdf encompassing both everyday fluctuations and “once in a century” fluctuations. Instead of a Gaussian or some correction to a Gaussian, they found a power law pdf with exponent -4, a sufficiently large exponent that the difference from a Gaussian is not huge; however, the probability of a “once in a century” event of, say, 100 standard deviations is exp(-10,000) for the Gaussian, but simply 10-8 for an inverse quartic law. If one analyzes a data set containing 200 million data in two years, this means there are only two such events–in two years!

Now which is better, the concept of “everyday fluctuations” which can be modeled with a drunkard’s walk, complemented by a few “once in a century” outliers? Or a single empirical law with no outliers but for which a complete theory does not exist despite promising progress by Xavier Gabaix of NYU’s Stern School of Management and his collaborators? Here we come to one of the most salient differences between traditional economics and the econophysicists: economists are hesitant to put much stock in laws that have no coherent and complete theory supporting them, while physicists cannot afford this reluctance. There are so many phenomena we do not understand. Indeed, many physics “laws” have proved useful long before any theoretical underpinning was developed . . . Newton’s laws and Coulomb’s law to name but two.

And all of us are loathe to accept even a well-documented empirical law that seems to go against our own everyday experience. For stock price fluctuations, we all experience calm periods of everyday fluctuations, punctuated by highly volatile periods that seem to cluster. So we would expect the pdf of stock price fluctuations to be bimodal, with a broad maximum centered around, say, 1-3 standard deviations and then a narrow peak centered around, say, 50 standard deviations. And it is easy to show that if we do not have access to “all the data” but instead sample only a small fraction of the 200 million data recently analyzed, then this everyday experience is perfectly correct, since the rare events are indeed rare and we barely recall those that are “large but not that large”.

The same is true for earthquakes: our everyday experience teaches us that small quakes are going on all the time but are barely noticeable except by those who work at seismic detection stations. And every so often occurs a “once in a century” truly horrific event, such as the famous San Francisco earthquake. Yet when seismic stations analyze all the data, they find not the bimodal distribution of everyday experience but rather a power law, the Gutenberg-Richter law, describing the number of earthquakes of a given magnitude.

Genomics of complex traits and evolutionary selection pressure on DNA regions linked to cognitive ability

Two recent papers relevant to the genetics of complex traits. The second paper is specifically about the genetic architecture of cognition. Thanks to blog readers for pointing them out to me.

See also The tipping point.

Genetics of complex traits: prediction of phenotype, identification of causal polymorphisms and genetic architecture

DOI: 10.1098/rspb.2016.0569
Proceedings of the Royal Society 27 July 2016 Volume 283, issue 1835

Complex or quantitative traits are important in medicine, agriculture and evolution, yet, until recently, few of the polymorphisms that cause variation in these traits were known. Genome-wide association studies (GWAS), based on the ability to assay thousands of single nucleotide polymorphisms (SNPs), have revolutionized our understanding of the genetics of complex traits. We advocate the analysis of GWAS data by a statistical method that fits all SNP effects simultaneously, assuming that these effects are drawn from a prior distribution. We illustrate how this method can be used to predict future phenotypes, to map and identify the causal mutations, and to study the genetic architecture of complex traits. The genetic architecture of complex traits is even more complex than previously thought: in almost every trait studied there are thousands of polymorphisms that explain genetic variation. Methods of predicting future phenotypes, collectively known as genomic selection or genomic prediction, have been widely adopted in livestock and crop breeding, leading to increased rates of genetic improvement.

Molecular genetic aetiology of general cognitive function is enriched in evolutionarily conserved regions

http://dx.doi.org/10.1101/063636

Differences in general cognitive function have been shown to be partly heritable and to show genetic correlations with a several psychiatric and physical disease states. However, to date few single nucleotide polymorphisms (SNPs) have demonstrated genome-wide significance, hampering efforts aimed at determining which genetic variants are most important for cognitive function and which regions drive the genetic associations between cognitive function and disease states. Here, we combine multiple large genome-wide association study (GWAS) data sets, from the CHARGE cognitive consortium and UK Biobank, to partition the genome into 52 functional annotations and an additional 10 annotations describing tissue-specific histone marks. Using stratified linkage disequilibrium score regression we show that, in two measures of cognitive function, SNPs associated with cognitive function cluster in regions of the genome that are under evolutionary negative selective pressure. These conserved regions contained ~2.6% of the SNPs from each GWAS but accounted for ~ 40% of the SNP-based heritability. The results suggest that the search for causal variants associated with cognitive function, and those variants that exert a pleiotropic effect between cognitive function and health, will be facilitated by examining these enriched regions.

World's fastest supercomputer: Sunway TaihuLight (41k nodes, 11M cores)

Jack Dongarra, professor at UT Knoxville, discusses the strengths and weaknesses of the Sunway TaihuLight, currently the world's fastest supercomputer. The fastest US supercomputer, Titan (#3 in the world), is at Oak Ridge National Lab, near UTK. More here and here.

MSU's latest HPC cluster would be ranked ~150 in the world.

Top 500 Supercomputers in the world

Sunway TaihuLight, a system developed by China’s National Research Center of Parallel Computer Engineering & Technology (NRCPC) and installed at the National Supercomputing Center in Wuxi, which is in China's Jiangsu province is the No. 1 system with 93 petaflop/s (Pflop/s) on the Linpack benchmark. The system has 40,960 nodes, each with one SW26010 processor for a combined total of 10,649,600 computing cores. Each SW26010 processor is composed of 4 MPEs, 4 CPEs, (a total of 260 cores), 4 Memory Controllers (MC), and a Network on Chip (NoC) connected to the System Interface (SI). Each of the four MPEs, CPEs, and MCs have access to 8GB of DDR3 memory. The system is based on processors exclusively designed and built in China. The Sunway TaihuLight is almost three times as fast and three times as efficient as Tianhe-2, the system it displaces in the number one spot. The peak power consumption under load (running the HPL benchmark) is at 15.371 MW or 6 Gflops/W. This allows the TaihuLight system to hold one of the top spots on the Green500 in terms of the Performance/Power metric. [ IIRC, these processors are inspired by the old Digital Alpha chips that I used to use... ]

...

The number of systems installed in China has increased dramatically to 167, compared to 109 on the last list. China is now at the No. 1 position as a user of HPC. Additionally, China now is at No. 1 position in the performance share thanks to the big contribution of the systems at No. 1 and No. 2.

The number of systems installed in the USA declines sharply and is now at 165 systems, down from from 199 in the previous list. This is the lowest number of systems installed in the U.S. since the list was started 23 years ago.

...

The U.S., the leading consumer of HPC systems since the inception of the TOP500 lists is now second for the first time after China with 165 of the 500 systems. China leads the systems and performance categories now thanks to the No.1 and No. 2 system and a surge in industrial and research installations registered over the last few years. The European share (105 systems compared to 107 last time) has fallen and is now lower than the dominant Asian share of 218 systems, up from 173 in November 2015.

Dominant countries in Asia are China with 167 systems (up from 109) and Japan with 29 systems (down from 37).

In Europe, Germany is the clear leader with 26 systems followed by France with 18 and the UK with 12 systems.

More genetic substructure

Via Genetic Future, two more principal components analyses of SNP data. Click the genetics label for related posts.

European, Nigerian and East Asian samples from HapMap and 100 African-American samples clustered based on data from ~600,000 genetic markers. The bend toward the East Asian cluster is probably due to Native American admixture.

Alkes L. Price, Nick Patterson, et al. (2008). Effects of cis and trans Genetic Ancestry on Gene Expression in African Americans, PLoS Genetics, 4 (12) DOI: 10.1371/journal.pgen.1000294.





Resolving Finns and Swedes. Jakkula et al. (2008) The Genome-wide Patterns of Variation Expose Significant Substructure in a Founder Population, The American Journal of Human Genetics, 83 (6), 787-794 DOI: 10.1016/j.ajhg.2008.11.005.

Genetic clustering: 40 years of progress

Represent each individual human by their DNA sequence. When aggregated, they cluster into readily identifiable groups. This has been known for 40 years now, although the technology and methods of analysis continue to improve. Below are results from 1966, 1978 and 2008.

If this seems counterintuitive to you, it might be because the space of genetic variation is of very high dimension. See here for more discussion and an illustration.

(Click images for larger version.)

Population Structure and Human Evolution
L. L. Cavalli-Sforza

Proceedings of the Royal Society of London. Series B, Biological Sciences, Vol. 164, No.995, A Symposium from Mendel's Factors to the Genetic Code (Mar. 22, 1966), pp. 362-379
http://www.jstor.org/stable/pdfplus/75457.pdf




Measurement of Differentiation: Reply to Lewontin, Powell, and Taylor
Jeffry B. Mitton

The American Naturalist, Vol. 112, No. 988 (Nov. - Dec., 1978), pp. 1142-1144
http://www.jstor.org/stable/2460361?origin=JSTOR-
pdf




Current state of the art, as discussed here. Figure: The three clusters shown below are European (top, green + red), Nigerian (light blue) and E. Asian (purple + blue).



According to the mathematical analysis given in this paper, populations with FST as low as .0001 can be resolved with current technology. (Typical FST between northern and southern Europe is about .006, between Europe and E. Asia about .1 and between Europe and Nigeria about .14 .)

Some favorite posts

NEW!  Podcast show Manifold.

I started writing this blog in 2004 (it has had millions of visitors!), and by now the content is a bit unwieldy to navigate, even with labels and search. I thought I'd make a list of some of my favorite posts and topics. Please suggest other posts to add to the list!

Pessimism of the Intellect, Optimism of the Will.

Feynman and me, Memories of Feynman. Labels: Feynman, path integrals.

Richard Feynman and the 19 year old me at my Caltech graduation:

Mama said knock you out (learning how to fight).

Many Worlds and quantum mechanics, a brief guide. Label: Many Worlds. Note, the usual quantum probabilities do not emerge naturally in this interpretation. See my papers On the Origin of Probability in Quantum Mechanics and The measure problem in no-collapse (many worlds) quantum mechanics.

Cognitive limitations: statistics , higher ed. Label: bounded rationality, human capital. Brainpower and globalization.

Expert predictions in soft subjects are unreliable. Intellectual honesty. Frauds! Label: expert predictions.

We can (crudely) measure cognitive ability using simple tests. (It is amazing to me that this is a controversial statement.) Randomly sampled eminent scientists have (very) high IQs, and given the observed stability of adult IQ the causality is clear: psychometrics works. The cult of genius? Income, Wealth, and IQ, One hundred thousand brains. Bezos on the Big Brains. Label: psychometrics.

Historically isolated groups of humans cluster genetically according to geographical ancestry. Explained in pictures , words , more words.

I am skeptical of all but the weakest claims of market efficiency. My talk on the 2008 credit crisis. Venn diagram for economics.

Careers, advice to geeks: A tale of two geeks , success vs ability. Labels: careers , startups , entrepreneurs.

Net worth , life satisfaction , happiness , the gilded age.

What is the likely development path for China in the next decades? Sustainability of China growth , China development: how big is the middle class? , Back to the future , Shanghai from an Indian perspective.

That curious institution, Caltech. How did a 16 year old kid from Iowa end up there? (See memories of Feynman above.)

There are geniuses in the world. The cult of genius.

My lovely kids. Photos. Autobiographical.


Update:

Credentialism and elite careers , Defining merit , elitism , brainpower

Recent videos (talks on genomics): https://www.youtube.com/results?search_query=hsu+genomics

Talks (some with slides + video):

Cold Spring Harbor Laboratory
Berkeley Innovative Genomics Institute and OpenAI
Janelia Research Campus (HHMI)
Allen Institute (Seattle) meeting on Genetics of Complex Traits

Review article: On the genetic architecture of cognitive ability and other quantitative traits (2014)

I work on algorithms for phenotype prediction from genotype, using new methods from high dimensional statistics. My estimate is that prediction of complex traits such as height, cognitive ability, or highly polygenic disease conditions will require data sets of order one million individuals (i.e., to build a model which accounts for most of the genetic variance). Once these models are available, human reproduction (and evolution!) will be revolutionized.

These papers are somewhat technical:
https://arxiv.org/abs/1310.2264
http://arxiv.org/abs/1408.6583

This one is a bit less technical and gives a broader overview:
http://arxiv.org/abs/1408.3421

Cow genomics (an existence proof):
http://infoproc.blogspot.com/2012/08/genomic-prediction-no-bull.html
http://infoproc.blogspot.com/2014/08/its-all-in-gene-cows.html

These are for popular audiences (Nautilus Magazine):
http://nautil.us/issue/18/genius/super_intelligent-humans-are-coming
http://nautil.us/issue/28/2050/dont-worry-smart-machines-will-take-us-with-them

2018: As anticipated, we now have good height predictors thanks to the 500k genome release of UK Biobank data: Scientists of Stature

Genomic predictors for common disease risk, constructed via machine learning on hundreds of thousands of genotypes. The predictors use anywhere from a few tens (e.g., 20 or 50) to thousands of SNPs to compute the risk PGS (Poly-Genic Score) for conditions such as diabetes, breast cancer, heart attack, and more: Genomic Prediction of Complex Disease Risk.

The Economist on polygenic risk scores (2019).

Detailed analysis of genetic architectures of disease risk predictors. Implications for pleiotropy.

Sibling validation of genomic predictors.

Recent papers from my group:

https://www.genetics.org/content/210/2/477
https://www.nature.com/articles/s41598-019-51258-x
https://www.nature.com/articles/s41598-020-68881-8
https://www.nature.com/articles/s41598-020-69927-7


2021 review article, prepared for the book Genomic Prediction of Complex Traits, Springer Nature series Methods in Molecular Biology: 

From Genotype to Phenotype: polygenic prediction of complex human traits