PRS and Paleoafricans

PRS scores for various traits, determined in Europeans, don’t work well for sub-Saharan Africans, due to big differences in population history.  That can in principle be fixed by doing GWAS studies among Africans.  But PRS scores determined in either Eurasians or Bantu surely won’t work with Bushmen, probably the most divergent human population.  And I can’t see how we fix this, because there aren’t enough Bushmen in the world to even do one of the required large-sample GWAS studies.

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

43 Responses to PRS and Paleoafricans

  1. Matt says:

    To the extent that the failure of polygenic scores to work across populations is not due to actual differences in causal structures but due to the non-causal “tag” SNPs not tagging the same variants in different populations, this is something that can, in principle, be fixed. However, there is bound to be actual causal heterogeneity as well.

    An ironic reversal we are already beginning to see is that “social justice” mongers shift from blank slatism to complaining that the benefits from genomic discovery, in terms of predicting disease and behavioral phenotypes, are flowing to Europeans while non-Europeans, especially Africans, are being left out.

    • Blank slatism never applied to health, I do not see this reflects any change. (or i mistunderstood you?) if polygenetic scores have poor predictive values for non-europeans, SJ guys’d hold that it signals of environment and oppression.

    • danodet says:

      I don’t get it. It Looks to me that this is indication that the PRS models are using only correlations.

  2. Steve Sailer says:

    Henry Harpending had lots of interesting anecdotes about Bushmen. For example, he recounted a time he accompanied a Bushman woman to a store. Apparently, Bushmen don’t have a word for a higher number than 3, but when the counter clerk tried to cheat the woman while making change, she correctly detected his subterfuge even though it involved numbers such as, IIRC, 72 and 28. Henry felt this was evidence against the Sapir-Whorf Hypothesis.

    • teageegeepea says:

      I’d only heard of the Piraha lacking numbers as high as you can count on one hand.

    • jb says:

      You wouldn’t have a link for that would you? Making change isn’t just a matter of being able to count (and additionally to add and subtract!), but also knowing the values of the various coins and bills, which would mean either memorizing those values or being able to read printed numerals. So I’d be interested in knowing the details of how the story played out. (Also, I thought the Sapir-Whorf Hypothesis, at least in its stronger forms, had been busted for quite some time).

      • syonredux says:

        “(Also, I thought the Sapir-Whorf Hypothesis, at least in its stronger forms, had been busted for quite some time).”

        Strong form, yeah. The weak form seems to hold up, though.

    • Toddy Cat says:

      Bushmen have IQ’s down almost out of sight, and many aspects of their behavior and situation confirm this, but the also seem to strike many observers like Dr. Harpending and others as quite clever, especially when it comes to thinking “outside the box” as they say. Are the Bushmen so different from us that our standard IQ tests are missing something? It’s not impossible…

      • The Z Blog says:

        Talk to anyone familiar with American ghettos and they will have stories about the remarkable cleverness of the inhabitants. At the same time, those same people will do things that are incredibly dumb. We know that the average IQ in these areas is quite low.

        This paradox is probably a form of observation bias.

        • Toddy Cat says:

          Do you think that this is also what is misleading Taleb?

          • gcochran9 says:

            I want to hear those stories. They might cheer me up. I mostly hear ones about some guy in Chicago who shot his wife after she broke off the light bulb in the socket by turning it the wrong way.

            • The Z Blog says:

              Here in Baltimore, the drug dealers will have very elaborate systems of lookouts to avoid getting caught up by the cops. The cops will tell you it is impossible to sneak up on them while slinging.

              They can also spot undercover cars. The reason is they have contacts in the garages where the unmarked cars are stored and serviced. The police have turned these repair facilities into bunkers in an effort to conceal the identity of the vehicles.

              In the city jail here, a well-known dealer was running his business from the cell. He was also impregnating guards, who would help facilitate his business. He was able to outwit the prison system, but he could not figure out how to work a condom, so the knocked up guards were his undoing.

              Is this cleverness or just a surprising lack of stupid behavior?

            • Kilo 4/11 says:

              Have some Chicago cheer:
              http://heyjackass.com/
              https://secondcitycop.blogspot.com/

              This Second City Cop thread from January https://www.blogger.com/comment.g?blogID=13350456&postID=8027677555495790602
              shows how the real police really feel about the neighborhood I lived my entire life in, up until last year (minus time in the military and some time spent seeing the world; but it was always the home I returned to).

              The closest I ever saw blacks get to being “clever” were the salvage guys who stripped condemned buildings of ornamentation, copper, radiators, etc., and the drug operations. Looking out my back door I’d see thirty-odd people materialize in the alley behind my house, form a single file, and wait patiently; several decoys would approach from different directions, crossing, passing through, and rotating around the line like a basketball team on offense eluding defenders, until suddenly one of them would begin handing out the drugs and taking payment, all in one motion. Never saw a supermarket checkout line move that fast.

          • The Z Blog says:

            I think Taleb is struggling to understand the concept of a smart fraction.

        • Ursiform says:

          Of the inhabitants, or of some of the inhabitants? Some clever inhabitants does not imply a generally high level of cleverness.

    • Anonymous says:

      Humans can broadly estimate quantity by instinct, so she didn’t have to rely on precise manipulation of numerals for a reality check. The clerk probably tried to cheat her by too much.

      Now I can recall a time when my lunch order was rung up by a mentally handicapped dude who was being trained (ineffectively) to make change. All he had to do was punch in the right numbers into the register and transact with the customer according to the result. I gave him a $20 for a $7.xx order, and he attempted to give me $7.xx in change, because both $7.xx and $12.yy appeared on his display and he was unable to infer which quantity was relevant. His handler had to intervene. The Bushman woman would outsmart THAT guy, but that’s roughly where the bar for these things lies.

    • JP Irwin says:

      uhhh hate to be the one to tell you but it’s 2019 buddy. “Bushperson woman” is the preferred nomenclature.

  3. Lauro says:

    Once we have PRS scores for sub-Saharan africans, we can compare them with the PRS of Europeans. There is bound to be quite some overlap. Maybe they cluster in some areas on some chromosomes. Maybe that gives us some partial idea which one’s would be informative in Bushmen.

  4. Smithie says:

    I suppose the attempt to make super Bushmen will have to be a trial-and-error process.

  5. dearieme says:

    “But PRS scores … surely won’t work with Bushmen”: why would I care? What practical purpose would be served by knowing this unknowable stuff?

    • Better Model says:

      I half suspect the main post is the usual Cochran “How obvious do we have to make it before everyone agrees with me?” thing, but there are actual implications that we could have better function models of human brain function with increasingly general genetic models with incorporate more human diversity.

      It would be even more useful to have a million chimps to try and get a nice little GWAS from them, but I don’t see that happening either.

  6. Capra Internetensis says:

    There must be a pretty large population worth of Bushman IBD tracts in Southern African Bantu and Coloured, though obviously interacting with mostly non-Bushman variants in rest of genome. How much could you figure out, given enough people who are a quarter Bushman?

    • dearieme says:

      Anyone who has seen a lot of white South Africans against a background of genuinely white people would guess that quite a few of them – especially the ones with Boer surnames – carry some Bushman/Hottentot DNA.

      • gcochran9 says:

        Well, I known the numbers – something like 5% if memory serves – and personally I’m not very good at noticing small admixtures like that.

        • dearieme says:

          If it averaged 5% then some Boers will carry 10% or more. Depending on the flukes of inheritance that could be easily visible to the naked eye. Maybe not to the average American naked eye, used to a minority of obviously black descent, perhaps. But obvious to my teenage naked eye, used to no blacks at all.

          It was seeing a Springboks touring party, under grey skies and against a background of genuinely white people, that made me realise that there was fair bit of hypocrisy in the race-theory aspect of Apartheid.

          Anyway, if 1/1024 can qualify an American lawyer to get jobs on a racial privilege argument, I’d think that 5% should give the Boers a jolly good crack at advancing in American society, should they ever gain asylum there. (Which presumably they won’t.)

          • gcochran9 says:

            If you’re talking about Elizabeth Warren, not 1/1024: 1-2%. For those that care about accuracy. Like me !

            If the admixture in Boers happened early enough, there may not be much variation in admixture percentage today. I don’t know the numbers off the top of my head.

      • Bob says:

        I know what you mean. Someone like Eugene Terre’Blanche, the late Boer nationalist leader, has the kind of phenotype that pops up among Boers and that one might attribute to Bushman/Hottentot DNA. Although it’s hard to tell if this sort of phenotype is genuinely more common among Boers due to African DNA, or if it’s within the typical range of European phenotypes and there’s just a bias to attribute it to African DNA as a result of the Boers’ connection to Africa.

      • Unconcerned Bystander says:

        Go on, then. Which ones? Plenty of Boeremens in this Free State Cheetahs squad. Tell us which ones you think have substantial Bushman DNA: http://www.fscheetahs.co.za/players

      • Unconcerned Bystander says:

        Go on, then. Pick a Currie Cup team and look at the squad online. Plenty of Boeremens in any of them. Tell us who you think has substantial Bushman DNA.

      • JP Irwin says:

        They are also admixed with Indonesians, such a Paul Kruger. That may account of some of the uncommon phenotypes

  7. Jacob says:

    Whites may have a higher frequency of the alleles that make Africans smart, much like Piffer’s findings that Asians have a higher frequency of most alleles that make Whites smart.

    “Egalitarians” aren’t running GWAS/IQ studies on Africans because they know they might falsify their own purported ideas.

    • Anonymous says:

      Indeed, there may be certain alleles that make Africans smart and are close to fixation in whites.

      • Jacob says:

        Not only does that hurt to read, but worse, I don’t even think it’s a bad guess. If they reached fixation, that would explain why they show up in their PRS and not ours: they didn’t account for variance in the population in which they’re nigh ubiquitous.

  8. a-non says:

    Something I find confusing is that the GWAS samples needed to discern traits seem large compared to the sizes of populations in which those trait evolved.

    What am I missing here, or can anyone recommend what I should read to understand such things? (As a numerate person who has never worked on anything close to this.)

    • Eponymous says:

      Evolution takes place over many generations, so the sample size natural selection is using to identify the +variants is much larger than the population alive at any one time.

      • a-non says:

        I guess this must be the core of it. But 1000 years only gets you a factor 300, and survival is a much noisier measure than (say) height in cm… although I guess the GWAS years-in-school is also a pretty noisy measure. I mean it must work out, I’d just like to have a better handle on the numbers.

        • dux.ie says:

          From Lee’s dataset part of the 1 million data points.

          The signals from the SJW rather than noise weaken the GWAS results. “No child left behind”, thus a horizontal cluster around 12 years irrespectively of ability. There might be social good that most citizens have at least high school education. But university education carries hefty student loan debts. AA, diversity policies and high university graduation rates, another higher horizontal cluster around 16 years for university graduates. But there is definitely on average a slope linking the centroids of the two clusters with respect to PGS. SJW’s self fulfilling assertion that genetics are not important. If there is no natural effects why the clustering around years 12 and 16? The fantasy faces real life when looking at the data from the Federal Reserve the under employment rate (graduates working in jobs that do no require ANY university degrees),

          http://tinypic.com/view.php?pic=r9onc3&s=9#.XF-DiIRxWRs

          and the early salary wrt the university entry IQ

          where in the SJW doctrine dominated majors with entry IQs less than 115 statistically the salary levels are independent of IQ as asserted by the SJW. Many in those clusters might be having problems repay the student loans. The picture is different in region IQ > 115. That is the real life data.

    • JP Irwin says:

      the problem is you’re trying to reveal, statistically, the tiny effects of many, many variables (the variants). You need big numbers so that the small effects can be revealed against the backdrop of the other effects as well as the non-deterministic component of the trait in question (called “nurture” sometimes, but better thought of as a white-noise residual). In an analysis like this, you want many times more observations (individuals) than independent variables (variants). In the case of the Bushpeople, you have far more genes to study than people.

  9. caethan says:

    C’mon Greg, you’re thinking too small. Start a war, push out the Tswana, rename Botswana to Hottentotopia, and start subsidizing 14 year olds having babies. There’s what, 50K of them now? If you can manage a growth rate of 10% a year, you’re up to 5 million in 50 years and Bob’s your uncle.

    We’ve certainly had stupider reasons for starting wars. At least we’d get something out of this one.

    Or if that’s a step too far, let it be known that you’ll pay good money to women who can prove with genetic testing that their child was fathered by a Bushman.

  10. Phille says:

    Could we do a Neanderthal PRS? I read 40% of their DNA is still around in current populations. And there are lots and lots of people with neanderthal DNA.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s