Yesterday I posted a link to an article in the UK Guardian, “The genes that build America” in which the author attempted to summarize some of the recent controversial topics in genealogical research, including DNA testing.
For at least one of my readers, the article represented everything that is wrong with DNA testing, specifically the assignment of racial/ethnic percentages based on the results of autosomal testing.
In the past, I’ve tried to be as impartial as possible when discussing autosomal testing. As I’ve learned, however, being impartial can also be unfair and misleading. So, I’ve decided to get a little more personal and share my thoughts about autosomal testing.
In a single sentence, autosomal testing is simply too new and underdeveloped to be of much informative use for genealogists or the average public, at least in its current stage. This statement, I hope, will be completely incorrect in a few years as whole genome sequencing becomes affordable. Assigning percentages (as autosomal tests do) will only work when the entire genome can be sequenced and examined and analyzed. Short of whole genome sequencing (and maybe comprehensive SNP testing - as in millions of SNPs), I don’t believe that autosomal is worth the effort.
The article, like most related articles, makes the single biggest mistake regarding autosomal testing - the belief that autosomal test results translate into ancestral percentages. For example, in the article Professor Peter Fine at Florida Atlantic University undergoes autosomal testing, and the results show that 25% of the markers tested are of Native American origin (For the purposes of this post, I won’t get into the debate over whether DNA sequences can be assigned to a particular ethnic group). The author of the article states that Professor Fine “discovered that he was a quarter Native American.” This, of course, is not the case. How can the sequencing of a few hundred bases of DNA reveal that 25% of a person’s entire genome, or their ancestry, is Native American? It can’t, not until a vast majority (or all) of the genome is sequenced and analyzed.
This mistaken interpretation of results can and does have devastating effects. Many science bloggers get upset when the media or even other scientists use confusing or incorrect terminology. Here, however, the incorrect wording is more than a simple misuse of a word - it has much larger implications, suggesting that autosomal tests can truly define who we are and where we came from. The numbers only represent a few hundred sequences bases out of an entire genome, and I do not believe that even full-genome analysis - the entire description of our DNA - will ever define who we are or where we come from, on an individual level. If I were to find a sample of my great-great-grandmother’s DNA and analyze her entire genome, I could tell you what color her eyes were, about how tall she was, and what diseases she might have suffered from. I could not, however, tell you what made her laugh, or what her hobbies were, or even how many children she had in her lifetime.
For an example of the impact that the misuse of autosomal testing can have on individuals, take Isaac Carter from the article. Mr. Carter’s family lore contains stories of Native American ancestry, but he was dismayed when his autosomal test revealed that he “had no Native American blood at all.” First, I don’t understand what the term “to possess blood” means - does it mean ancestry, or does it mean DNA? Second, since the test was just a tiny, miniscule fraction of the entire genome, the results do NOT rule out the existence of sequences of DNA with Native American origin. And third, the lack of Native American DNA does NOT rule out the existence of Native American ancestors! I believe, and I have always believed, that even if we do not possess a single base of DNA from an ancestor, we are still their descendant - our very existence is the result of every single decision that they made, and we are their legacy, regardless of how much of their DNA we can identify. Mr. Carter, don’t be dismayed by the results of this test. Instead, use the curiosity that it has generated to do more research and learn more about your “paper trail genealogy.”
No autosomal test, now or in the future, will ever be able to completely define a person or a person’s heritage. DNA inheritance simply doesn’t work that way; it happens by chance, and as a result we do NOT possess DNA from each of our ancestors evenly. Some are highly represented, and some are gone completely. The fact that autosomal results suggest that 25% of the markers tested are of Native American origin does NOT mean that 25% of that person’s ancestors were Native American. And this is the most confusing aspect of autosomal testing.
The article is also discussed over at SciGuy.
6 Comments
What do you think of DNA Tribes? They offer autosomal DNA testing that does not calculate percentages and instead relies on “top population matches.”
And when it comes to ethnicity DNA testing, how reliable are results from mtDNA and Y-DNA testing?
Great questions as always Hsien! The percentages are really what make autosomal tests so confusing. I think that ‘top population matches’ is a better way to go, although I still see lots of people that are confused by the results.
I put MUCH more stock in assigning geographical origins to Y-DNA and mtDNA results. They’re passed down almost completely unchanged, and there’s been so much more research to establish the origins of Y-DNA and mtDNA. Some of the most convincing support is research that maps ancient mutations in Y-DNA and mtDNA as it traveled from Africa to its final location. Autosomal DNA just doesn’t have enough of that type of research yet.
Note that there’s a handy ethnic origins comparison chart at ISOGG.
Blaine,
We’re 100% in agreement that autosomal DNA testing for finding individual admixture proportions is not ready for prime time yet.
But I don’t quite agree that we’ll need full genome sequencing before these tests can be made to work.
We just have to keep in mind that conclusions based on sampling always come with statistical uncertainty.
You are entirely correct that a couple hundred SNPs (of which only a subset are chosen to distinguish between Europeans and Amerindians) can’t tell someone he is “25% Native American”. 25% is the “Maximum Likelihood Estimate” calculated by DNAprint’s algorithm, but chances are fleetingly small this is his true admixture proportion. Even if the test is free of systematic error and DNAprint’s marker selection and population reference samples are adequate (they’re not — see below), DNAprint reports results with wide confidence intervals (for example, this art teacher’s “Native American” result may be something like 25% +/- 10% at a given confidence level).
For someone trying to prove an Amerindian ancestor 4+ generations ago (this seems to be one of the typical DNAprint applications), the built-in 8-10% margin of error renders the results meaningless, before we even look at any other problems with the test. The test is even worse for people who discover “admixture” they had no reason to suspect; these people typically waste time grasping for explanations for their “exotic” ancestry instead of recognizing it for the statistical noise it most likely is, judging from messages I see people post on various boards and mailing lists.
Going back to the art teacher: as I recall, his name is “Fine”. I’d give better than even odds he is of Jewish descent. DNAprint’s website notes “low levels of Native American affiliation are commonly observed for Jews [using DNAprint’s tests].” DNAprint goes on to try to rationalize the finding by implying that the “NAM affiliation” the company detects among Jews, Middle Easterners, and Southern Europeans is “real” and the result of common ancient ancestry.
Obviously, all humans — all living things for that matter — share common ancestry if we go back far enough. People aren’t paying hundreds of dollars for DNAprint to tell them that.
In fact, these anomalous “NAM” results are prima facie evidence DNAprint needs to use more markers and add more “parental population” samples to their databases. But DNAprint exists to make money. They have no incentive to improve as long as they can keep duping customers with their current offerings, helped along by scientifically-illiterate journalists giving them uncritical free advertising.
Hsien,
I’d recommend you read the discussion about DNA Tribes on the DNA-GENEALOGY mailing list. In particular, I suggest you pay attention to comments by John Chandler (one of the sharpest posters on the list, who has also made good posts about DNAprint’s tests). This Google search should get you started.
xc,
I would just like to thank you for participating in this discussion, and adding your thoughts. It makes the topic much more interesting. I hope you stick around and join in often. Blaine
I have a few comments about DNA Tribes. My mother is French an my father was Kabyle (berber). I did the test in May and my results showed that my profile was typically european, very close to Swiss people (very coherent with my mother side). At this time they used only 13 markers for the test. Last week they added 2 new markers Penta D and E and after I updated my results the new report showed that this time I was very close to Tunisian people and Switzerland had now a extremely low score. So it is very confusing cause the results were completely different which means that the test in my opinion is not very reliable for mixed people
I have signed up to take the autosomal test Euro DNA through Genetic Testing Laboratories, but I have since found alot of the discussions about the inaccuracies in these autosomal tests. I aleady know that my line is European, but I thought the test would be interesting. Now I am concerned that I have wasted my money. I just wonder whether I should take the results in any serious light.
5 Trackbacks
[…] Bettinger, our favourite genetic genealogist, describes the pros and cons of autosomal genetic testing. He says, no autosomal test, now or in the future, will ever be able to completely define a person […]
[…] Bettinger, our favourite genetic genealogist, describes the pros and cons of autosomal genetic testing. He says, no autosomal test, now or in the future, will ever be able to completely define a person […]
[…] always been my belief that personal genetics (inexpensive whole-genome analysis) will bring about some […]