Dubious parentage

It would seem like sweet poetic justice if James Watson were found to be 1/8th African, but I’m afraid I don’t quite believe it. This is news coming from a company called deCODE genetics, an Icelandic outfit that analyzes an individual’s racial background on the basis of various genetic markers. While I can buy the claim that they can assess the distribution of various alleles in populations, I really dislike the game of trying to work in reverse and assign the fraction of a race to an individual.

I don’t think Larry is much impressed with them, either. Here’s another article that brings up the flaws.

The problem is compounded by the increasing improbability of the company’s claims. Watson is also 9% Asian. He’s got a 31% lesser chance than average of getting asthma, and a 2% greater chance of prostate cancer. These kinds of numbers are meaningless when applied to individuals. We don’t even know all the genetic factors that contribute to the various diseases listed, so it’s ludicrous to pretend they can quantify the total risk for a single person that way. While I’ve got no problem thinking there are shared alleles percolating through African and European populations, there are much more reliable ways of determining that a person has an African great grandparent.

(via Accidental Blogger)


  1. Sigmund says

    I can’t seem to find the details of how this result was determined. Is it an average of european asian and african SNPs or did they use find 16% of his genome consisted of african haplotypes?

  2. says

    These kinds of numbers are meaningless when applied to individuals.

    I assume they mean “people with this haplotype have been observed to develop asthma 31% less often,” which is meaningful. It could still be a poor predictor for polygenic traits, though, especially with admixture (like they’re saying happened here). Regardless, I’d like to see error bars for this and for the ancestry.

    I can only imagine how much more confusing this would be to someone who isn’t a geneticist. Maybe that’s why you need a license to interpret most other medical diagnostics.

    there are much more reliable ways of determining that a person has an African great grandparent

    I’m not so sure. Jerry Springer’s existence is evidence to the contrary. It’s not outside the realm of possibility that, well over a hundred years ago (Wikipedia says Watson was born in 1928, so extrapolate), there was a case of mistaken paternity and the real father was black. I don’t even know most of my great-grandparents’ names, and I do know something like that would be swept under the rug in my family.

    Aside from the irony, though, what difference does it really make?

  3. says

    >>These kinds of numbers are meaningless when applied to individuals.<< Uhh, no they're not. I presume they are the probabilities calculated from some model. They may not be particularly conclusive but they're not meaningless.

  4. Janek says

    There are DNA testing companies which offer genuine genealogical value, in that they can confirm whether or not two people share a common paternal ancestor.

    Companies which claim to sell you your “ethnicity” (whatever that means) aren’t offering value for your dollar, IMO.

  5. D says

    It’s also the self-undermining argument. To say that you can actually determine the race of a person’s great-grandparent is certainly to open up the possibility that genetic admixture’s effects on phenotypic traits can be measured.

  6. John C. Randolph says

    This sounds a lot like the BS genealogy services that used to advertise by direct mail in the 1970’s.


  7. Karen says

    Aww, bummer – my dad and his siblings are adopted, and I have been toying with the idea of running the whole family through something like the National Geographic Genographic project. I had thought the unveiling would be a fun family event as well as providing something resembling ancient heritage in the family. Noone has ever searched out biological family, but we had thought it would be interesting to get an idea of where our ancestors trod.

    As I understood it, they look for marker genes that developed in particular populations – are the waters still all that muddy if you’re interested in broad strokes?

  8. Steve LaBonne says

    Money quote from that second link PZ gave:

    “That’s the beauty of this scam. The companies aren’t scamming you. They’re not giving you fraudulent information. They are giving you data, real data, and allowing you to scam yourself.”

    Exactly- that’s what makes this an especially insidious form of flimlam.

    As a forensic biologist I’m just tickled to death- not- that any day now cops are going to start asking about this stuff…

  9. Julie Stahlhut says

    Back in the old days when genotyping was done via starch-gel electrophoresis of enzyme variants, I was reading a paper on the subject that included a table of human population data. One locus immediately caught my eye, because it was at near-fixation for one allele in Italy (hi, Mom!) and a different allele in Poland (yo, Dad!) I joked with my friends that if I ever went missing, they now knew exactly which heterozygote to look for.

    Regardless of the markers used, though, I wouldn’t expect to be able to run the numbers backwards. And it’s oversimplistic to assume that a person whose genotype is consistent with being one-eighth Grand Fenwickian got that way because of a single Grand Fenwickian great-grandparent.

  10. says

    This blogger has a good primer on how the DNA ancestry testing is done. He’s against Small’s contention that it’s bad science and says so in no uncertain terms.

    Perhaps we should just go about using the method described on Dr.Watson’s published genome to see if we can independently come up with results proving or disproving the assertions in the article ;)

  11. molecanthro says

    cops are already asking for this information. there have been multiple cases now where they are using companies to identify ‘ethnic’ origins. check out this story
    and they even have tests for eye color. we had someone come to our institute a few weeks ago from the Department of Forensic Molecular Biology at the Erasmus University Medical Center to give a talk on eye color genetics.

    police are using this information to narrow down suspects and build a case from there. they aren’t using this information in court like they would with ‘dna fingerprinting.’

  12. Steve LaBonne says

    Sujatha, read the comment thread on Larry’s blog post where this was dealt with. Your blogger is missing the point. At least some of these companies, while obtaining results that are indeed scientifically correct, are allowing / encouraging customers to vastly overinterpret what the results mean. Here’s a lovely example of advertising copy dug up by Larry:

    With a simple swab of your mouth, you can be compared with Marie Antoinette and discover your relation to her. You will be able to trace your heritage back to times of grandeur, to times of kings and queens and royal balls, and possibly even to Marie Antoinette.

  13. Steve the Skeptical DoDo says

    Is that to say that the work of Spencer Wells with the National Geographic institutes genographic project is so much hogwash? I read his book and his methods and results in trying to trace roots, but not precise particulars on race, eye color, etc., was valid.

    Steve the Skeptical DoDo

  14. Brigit says

    I wonder what an analysis of my DNA to determine ethnicity would pop out. I’m from the Caribbean and according to accounts from my families I’ve got Spaniard, African, Taino, and Swedish ancestry (and who knows what else).

    Some adventurous Swede family moved form Sweden to Jamaica, then emigrated to St. Croix, and finally ended up in Puerto Rico. They had plantations and slaves and some serious cash.
    My branch of the family got disinherited when my greatgreat(great?)-grandfather challenged his dad because he gave his last name to his sons born by a slave. My G^3grampa changed the last name somewhat and moved out of town. One of his kids married my creole G^2grandma who had both indigenous and black ancestry (suck it, G^3grandpa!).

    From what I can gather, my other ancestors were creole peasants from the mountains and slaves from the Loiza settlement. One of them was incarcerated by the Spaniards in one of the forts (now a US museum) for rebelling.

  15. says


    I make no claims for the blogger I link to, beyond the clarity of the description of how to determine the haplotypes in one’s DNA and how to match them up against the predominant groups by geographic location.
    Advertisement is what it is, they will say anything they wish to make a buck or two (witness the ‘protect your child at birth by banking their cord blood’ advertisement, as I mentioned in one of my posts on A.B.)
    I agree that the buyer of these services should beware, but that doesn’t negate the science behind it.

  16. DiscoveredJoys says

    Although I agree that there is a temptation to over interpret these tests, the Live Science article makes just as big a sweeping assertion the other way…

    “Humans have, in fact, turned the whole world into one large genetic melting pot. We have always been a species that crossed mountains, continents and oceans; we have always loved to mate outside our ancestral group.”

    There is obviously some truth in this view, but it is overstated. We collectively are not all the same build, skin and eye colour, or blood group.

    It is only recently (in evolutionary terms) that wide travel has increased the spread of genes beyond slow migration speeds. Some people would like to avoid the social and genetic arguments about ‘race’ – which I agree is a poor label – yet others have identified geographic ‘genetic clusters’. For instance people living at high altitude in South America and Tibet show genetic changes reflecting their low oxygen environment – but the two demes have different genetic changes.

    There have been some very heated blog debates about ‘race’ and ‘intelligence’ recently. I am still waiting to see the ‘proof’ that no such difference *can* exist. I would be quite happy for such proof to be provided, but I am not willing to accept the assertion merely because it is comfortable to believe so.

  17. robotaholic says

    Is it true that there exist differences between human races? the answer is YES, is it possible that intelligence could be one of them? the answer is YES. Could he be right? YES. I am tired of this story as it no more proves or disproves his theory. If he turns out to be right, I laugh in the face of those who got on their HIGH HORSE to defame this man -you too PZ-(honestly I was surprised at your response)- if he turns out to be wrong, at least he said something falsifiable and therefore scientific. Just because something is comfortable doesn’t make it true..(I beleive Richard Dawkins uses that line alot when dealing with the overly emotional.)

  18. rhian says

    OK, I haven’t read anything about this, and so I don’t yet have an opinion on whether these conclusions are valid.

    But calling deCODE an “outfit that analyzes an individual’s racial background on the basis of various genetic markers” is completely deceptive. Yes, it’s a company, but certainly not even close to those who sell you your “racial” background for several hundred dollars. deCODE is actually consistently on the forefront of much of genomics, especially in finding new risk variants associated with complex disease. Kari Stefansson is no shyster scientist, and I’m really surprised that you imply as much. To be honest it sounds a lot like you’re putting your politics before your science.

  19. says

    Racial admixture testing is currently reasonably reliable for large groups of people, but can give dubious results for particular individuals. And this sounds like one of them.

    The first half dozen pages of Watson’s autobiography “Avoid Boring People” show pictures of his father, mother, and maternal grandmother and gives detailed information on his family tree and relations. His maternal ancestors were fairly recent immigrants from Ireland and Scotland, so if they were 100% white, then Watson’s father would have to have been only 50% white, which seems highly doubtful. Since the Watsons came from a socially fashionable family (the scientist’s grandfather was a stockbroker and his grandmother an heiress). America was too racist a century ago for a half nonwhite clan to have flourished the way the Watsons did.

  20. ovald says

    I don’t know PZ Myers an associate professor at The University of Minnesota, Morris, nor do I know K. Stefansson, CEO of deCODE genetics. If I did know you better, I am sure that I could call you a scammer. I just hope for your sake that you do not know Mr. Stefansson from previous university life?

    PZ Myers doesn’t do his homework, nor does he seem to do any research? That is why I am surprised that you have done this write-up without getting your facts correct, you do of course refer to “Larry article” but your mistake is not knowing much about deCODE! For an associate professor to call them a scam is ludicrous to me. There are two very good blog articles here by molecanthro and rhian. It looks like rhian has hit the nail on your head?

  21. spurge says

    @ ovald

    Why don’t you tell us all why PZ and the others are wrong in their assessment of deCODE?

    unsupported assertions that they are wrong don’t count for much.

  22. says

    I was tempted to agree with PZ on this, but what are the odds?

    Look, if the Independent and the Times say that Watson is black, then he’s black–by definition. Read the papers. The evidence keeps mounting with every new story. Can all of these objective reporters and publishers be wrong?

  23. Stevie_C says

    Sounds like the Icelanders don’t like getting a little hear for their press grabbing goofiness.

  24. rhian says

    Can we please stop confusing ancestry (genetic) with race (social)? No one is saying that Watson is black. Jesus.

    As for Watson’s “half nonwhite” clan, there is no reason to believe that ancestry should correlate to what we classify as observable racial phenotypes. There are many multiracial people–even with a full 50% of “nonwhite” ancestry, including myself–who pass so well for white that no one thinks twice about it.

    No one here has read anything about the actual scientific analysis, because there is nothing written about it. We don’t know how they got these percentages, we don’t know anything about the levels of error, or the distribution expected in a homogeneous population. The best I’ve heard is “he has 16 times the number of genes considered to be of African origin than the average white European does” from the NYTimes, and that already smacks of journalistic misinterpretation.

    Healthy skepticism is great, but maybe we should all stop jumping to conclusions before we see the data. I think that’s what most people here are all about anyway.

  25. says

    I don’t know how exactly how deCode did it, but Watson’s genome is publicly available and his SNP content is downloadable. He certainly has several alleles that are found in the YRI (Yoruba) population that are found at < 1% in the CEU (Central European - Utah) or other populations studied in the HapMap. For example, go to his genome browser and search for rs3922960. There are many cases like this that are easily found. Translating that into possible genealogies is something I’m less sure about.

  26. says

    ugh.. shouldn’t try to comment right before a faculty meeting. Should read:
    ” He certainly has several alleles that are found in the YRI (Yoruba) population that are very rare in the CEU population”

    Actually I think I wrote something like that but my attempt to put a < 1% truncated the comment.