How much science in genealogical DNA tests?

August 3, 2009 3:50 PM

Help me, MeFi geneticists! Commercial genealogy testing - all it seems?

I've got a few questions about genealogical DNA testing - you know, the sort that purports to tell you about your ancestors in exchange for a large but affordable sum. This has been provoked in part by a relative of mine going in for one of the tests, and being very disappointed in the result, or at least as much of it as she understood. I've had a look too, and I can't work out how to check the promises by the companies nor, given that they're often run by academics still in place in universities, the academic nature of the science they use.

These tests apparently look for markers for single nucleotide polymorphisms on the y chromosome or mitochondrial DNA, places where there's variation between populations in humans. Supposedly these can show you shared a common ancestor with groups. But can they do all that's claimed - when a gene came into a country, or when a mutation occured, and if so to what precision? Can they identify a gene as belonging to group such as Irish, or Celtic, or whatever? And, perhaps most pressingly, how much of the promise of the commercial personal genealogical DNA testing industry is based on academically respectable and applicable work? Some of these places seem to offer quite detailed breakdowns of when and where a particular marker became important, and I don't understand how.

Any enlightenment or pointers thereto, much appreciated!

(I''ve read the previous Ask MeFi posts on this, but they don't seem to cover the above)

posted by Devonian to science & nature (10 comments total)

I did the National Geographic Genographic Project one years ago. I found out that my ancestor(s) through my mitochondrial line (as I'm female) were part of a group that was in Europe 50,000 years ago. It's not that specific. It doesn't tell you so specifically as far as ethnic heritage. My white boyfriend did it too and his y chromosome traces back to a haplotype found x thousands of years ago in India. He is of almost completely English descent.
One feature is that you get emailed when someone matches you, and you can see what country those people are from. Most Americans, like myself, don't know what ALL of their ancestry, just a guess based on where their known ancestors came from, just some. So unless you're fairly certain of your ancestry, it's a bit of guessing when you fill out the form for the matching.
Honestly, looking back, it seems silly because you're finding out what large group ONE of your many ancestors 50,000 years ago belonged to.

Maybe other testing ones are more specific, but not the one I did.
posted by ishotjr at 4:24 PM on August 3

Also, the only part of your question that tries to answer is about finding out if one is Irish, Celtic, what have you. It can only tell you how common that haplotype seems to be among current ethnic groups. For example, mine is found in about 10% of Europeans, across all European groups. I'm clearly of mostly European descent by virtue of being a white blue-eyed person, so it doesn't tell me anything about where in Europe my ancestors might have come from at all.
posted by ishotjr at 4:27 PM on August 3

It really only tells you about one particular line.
posted by kldickson at 4:30 PM on August 3

WNYC's excellent show Radio Lab did a show on Race late last year, one segment of which was on these swab & mail DNA profiling services. You may want to give it a listen (it's a free download). If I remember right, the results only gave you results as percentages in four categories: European, African, Native American, and Asian, so no, it didn't even make a distinction between, say, Greek and Scandanavian.

Things may have progressed a bit since then, or maybe other companies are more credible, but it sounded like a bit of a vanity scam to me.
posted by Ufez Jones at 4:33 PM on August 3

FamilyTree DNA has a SNP assurance program that'll run on Y chromosomes. They run a prediction based on your numbers from the regular test and if the confidence isn't 100% they'll do the "Deep Haplotype Test" which'll give you a really ideal of ancient ancestry (thousands of years ago). In my case I'm m222+ (Ireland/Scotland).

Now here's where things get murky. I'm not a dead on match for m222 and these things are changing all the time. Note the tree for just the R tree here. It's dated for a reason. Later they might decide a group of us all match each other and declare a sub tree of m222, or a parent/sibling tree, etc. In the mean time we're left wondering how we got to County Donegal and where we went from there. In my particular case I have to address how we got from there to Sweden and finally over to the states, where despite my freckles and red hair the family believes we're from Sweden and has the genealogy black holes (pre 1880 is a complete mystery... long story) to back it up.

Before the DNA testing I just had the black holes. Now I just have more interesting black holes. I'm not sure if that is good or bad though. YMMV.
posted by jwells at 6:11 PM on August 3

Y-chromosomes are inherited from father to son Y chromosome DNA tells you about your father, his father, his father etc. This means that going back 10 generations, Y-chromosome DNA can only tell you about ONE of your 512 great great great great great great great great grandfathers. Similarly, mitochondrial DNA is inherited only from your mother and will tell you about one and only one female ancestor in each generation on your mother's side of the family.

Mitochondrial and Y markers are useful for studying generalised population movements, they are used for this purpose by reputable academcs and although I don't know the current state of the field, I'm sure the resolution of such mapping is getting better all the time. But this information is of extremely limited use in determining the genetic heritage of individuals. You learn nothing about the majority of your ancestors, just one man and one woman from each generation.
posted by nowonmai at 7:37 PM on August 3

I don't want to retype it all, but I wrote a big answer on this very subject in 2007 here. I run three surname-based DNA projects at FamilyTreeDNA.com and have turned into a true genetic genealogy fiend in the past few years, so this question is right up my alley!

This has been provoked in part by a relative of mine going in for one of the tests, and being very disappointed in the result, or at least as much of it as she understood.

You say she was disappointed? This is not surprising -- being female, she can only test her mitochondrial DNA (mtDNA), which is the direct maternal line. mtDNA mutates very slowly, so even if you have an exact match with someone, your common maternal ancestress may have been hundreds or even thousands of years ago. That being said, do you know if she went for the HVR1, the HVR2, or the full-out FGS test? Because the more markers you test, the better your results. Furthermore, has she looked for any matches in non-proprietary databases, like mitosearch.org or better yet GenBank? She might have better luck there.

I did the National Geographic Genographic Project one years ago. I found out that my ancestor(s) through my mitochondrial line (as I'm female) were part of a group that was in Europe 50,000 years ago. It's not that specific.

That's because unlike paid genetic genealogy firms like FamilyTreeDNA.com, the Genographic Project only tests HVR1, which is pretty worthless for recent genealogy purposes. They probably couldn't even tell you your haplogroup besides just plain "T" or "H" or something. Whereas if you did an FGS test (pricey!) you might learn that you're more specifically in T1a or H11 or something, and have two and only two exact matches. mtDNA has been used to prove genealogical relationships before, but it's pretty rare -- see the go-to story of the Kelly sisters for details.

(By the way, the story I mentioned in my 2007 AskMeFi comment linked above, where I thought I found Scottish heritage through a mtDNA test? Turned out it was a mislabeled sample in their lab. I'm actually in L2a1...which is a really rare and [so far] wholly Ashkenazic Jewish mitochondrial DNA haplogroup that hails from sub-Saharan Africa within the last 1500-2000 years, very likely as a result of Roman Empire era slavery. Great-great-...-great-grandma Cleopatra Schwartz, FTW! Alas, I am unlikely to find a paper trail of her existence, but it was still very cool to find out. So even mitochondrial DNA can be genealogically useful, but in a general sort of way.)

That being said, as a female genealogist, it pains me to say this, but given the state of technology at the present time, the real bang-for-your-buck value of genetic genealogy is in testing males. Since she lacks a y-chromosome, your relative needs to swab her brother or father or male cousin or nephew, or whatever male line she wants to start with. His y-chromosome is much more likely to give you good results, as y-chromosomes mutate much more quickly (on average, one marker value mutation accumulates after every 8 generations or so), so if you find a very close match to another male, then you and he can roughly calculate when your male ancestor lived, give or take two hundred years or so. If one of you has a paper-trail genealogy already that says "hey, my paternal ancestors were living in such-and-such county/parish/oblast during that very century", then you suddenly have a new lead to use for your genealogical searching.

Note too that you really need to have at least 37 or better yet 67 markers tested for your y-chromosome, too. Knowing that you and John Doe match on 21 out of 25 markers isn't very useful if you're both in a common Western European haplogroup like R1b1b. But if your match holds up to further marker testing and you still match at, say, 63 out of 67 markers, that is a big deal. The more markers you test, the better you can nail down the exact time period you likely shared an ancestor. More markers = more specificity.

Can they identify a gene as belonging to group such as Irish, or Celtic, or whatever?

Let's be clear on the terminology here: most genetic genealogy firms are not testing genes, exactly. They are counting the number of marker repeats on a chromosome, or STR's. They are also looking at SNP's, which are certain points on a family tree that totally mutated from an earlier version, which helps break up the human family tree into different branches. They are not looking at what those repeats and SNP's may or may not actually do in you, as genes. But they can let you know that the marker numbers they see represent certain "motifs" that are found more often in certain ethnic groups or areas.

Every branch of the human family tree, as traced through the y-chromosome, gets assigned a letter of the alphabet. Sub-groups then get a number. Sub-sub-groups get a letter again, and so on, as more and more finely delineated branches are discovered.

So for example, my brother is in y-chromosome haplogroup G2c. That means he's in haplogroup G, then got sorted into G2, and finally G2c. There are also branches G2a, G2b, etc., but he's not in those. There is even a further-down sub-haplogroup called G2a1, but he's not in that one; he's up from that a level because he lacks G2a1's particular defining SNP.

Now in his case, merely knowing he's in G2c does tell him about his ethnicity; something like 97% of people in G2c are known to have Ashkenazic Jewish ancestry. (Some testees' ancestors converted to Anabaptistism or other religions over the years, but the families retained oral histories of long-ago Judaism that were borne out after the testing.) There are people in G2a from Wales; there are people in G2c1 in Pakistan. That's a lot of geographic range for this trunk of the tree to cover over the years. Their particular branches help them, but his (my, if I were male) branch helps me.

Now, I already knew we were Jewish, but what was useful about doing genetic genealogy testing, to me, as someone who wants to find paper documentation (if possible), is that if you sketch out my brother's very closest paternal genetic matches on a map of Europe, most of them lived within 100-200 miles of L'viv, Ukraine a few hundred years ago. They were generally not living in Germany or anywhere in Western Europe. From e-mail conversations with the matches, there is no history of any rabbis or rabbinical lineages for any of the G2c's. This is odd and interesting to me, and could be clues for further research. They also all have totally different last names, which was expected since most Ashkenazic Jews only adopted surnames about 250 years ago, but now I know which surnames are potential relatives, and if I see them in a record I'm researching (like an early tax list for a town) it could be a clue. So in this case, genetic testing has given me an edge to my regular research...which is how genetic genealogy should be used, as an adjunct, not as a substitute for research!

but it sounded like a bit of a vanity scam to me.

No, there is serious genealogical work being done with these test results. I'm only an amateur genealogist, but I run three surname projects. In my Russo/Rousso project, we've managed to partially reconstruct a hundreds-of-years-old family tree for my mother-in-law's family, despite the lack of Ottoman Empire era paper records, thanks to creating phylograms based on marker values from many Russo/Rousso/Rousseau men around the world who signed up to join the project. In my Ganz project, we just reconnected an elderly Holocaust survivor in Australia with my husband in Los Angeles, proving that their Ganz families from northern Romania were once one and the same, despite local Romanian records not having survived for Jews before 1854. In my Schreier project, we just had the son of a man who was sent from the Czech Republic to rural Canada (adopted by a German-Canadian family) during early WWII send in his test kit, and we're awaiting results. See, he had three other siblings and his kids think his mother was Jewish and sent her four children off with strangers in four different directions in the hopes that at least some of them would survive the war, and right now that genetic test is the only way to prove this... His results are due back from the lab next month.

And I'm only an amateur; you should see what the professional genealogists are doing with this kind of new information...amazing stuff.

Or check out the Genealogy-DNA mailing list at Rootsweb, or the International Society of Genetic Genealogy and their "Success Stories" page.

Mitochondrial and Y markers are useful for studying generalised population movements, they are used for this purpose by reputable academcs and although I don't know the current state of the field, I'm sure the resolution of such mapping is getting better all the time. But this information is of extremely limited use in determining the genetic heritage of individuals. You learn nothing about the majority of your ancestors, just one man and one woman from each generation.

This is true. But if you're nutty enough to also test your other relatives -- grandparents, uncles, cousins, etc. -- you can learn about each of their direct maternal line and paternal lines, filling in the gaps in your own tree. And you can compare notes with other genealogists from distant family branches who swabbed their own relatives. Over time, you can build up information on a number of your ancestral "treetops". My son, for example, so far has had five of his direct ancestors tested, plus side branches like his uncle (my brother) and my husband's uncle and several of that uncle's cousins, who are in a Sephardic-focused genetic study. Ten years from now, when the technology has progressed a bit, he could have a nice amount of information available to him from those test swabs.

I honestly think genetic genealogy, and personalized genetic medicine, and DIY DNA are a huge new frontier. This is like the early days of the personal computer. It's gonna be huge.

By the way, I offer hope to my fellow female genealogists that X-chromosome testing and matching databases are coming...someday. The problem is that X-chromosomes recombine before they get passed onto you, unlike y-chromosomes that just get passed on as-is (except for the occasional small mutation). So if you're male, your one X-chromosome is potentially a mix of your mom's two X-chromosomes, one of which she got from her mom and one of which she got from her dad's mom. If you're female, you've got two to test, and it's just that much harder. But someday...
posted by Asparagirl at 11:54 PM on August 3 [4 favorites]

>> Devonian: "Help me, MeFi geneticists! Commercial genealogy testing - all it seems?
...I can't work out how to check the promises by the companies nor, given that they're often run by academics still in place in universities, the academic nature of the science they use.

Check out AABB accreditation. AABB accreditation is the gold standard for labs that do genealogical DNA testing.

Only a handful of labs out there do the DNA testing that almost all the genealogy services resell. For example, Ancestry.com resells the testing of Identigene. A couple other companies resell the testing of Sorenson Genomics. (Not sure who Family Tree DNA uses)

What differentiates the DNA genealogy test resellers is what they do with the illegible dots and numbers once it comes back from the lab. Someone has to translate that raw data into meaningful genealogy information that a layman can understand.

>> These tests apparently look for markers for single nucleotide polymorphisms on the y chromosome or mitochondrial DNA, places where there's variation between populations in humans.

There is also the third option of an autosomal test, which is a relatively new procedure and tests across one's whole genome, regardless of gender. With the autosomal test, the limitations of Y-chromosome and mtDNA testing become moot. This is a good test of you want to find out about what world populations you come from -- but says less about the heritage of your particular surname, for example, or who your oldest female relative is. It all depends on what your research goals are. Not many companies offer the autosomal test yet, and I know one of them that did has gone out of business (DNAPrint, I think?), but you should at least check it out when researching genealogical DNA testing. I've physically seen the results of an autosomal report, and thought that it did what it said on the tin, so to speak.

>> And, perhaps most pressingly, how much of the promise of the commercial personal genealogical DNA testing industry is based on academically respectable and applicable work?

A lot of it. It isn't all junk science or voodoo... there are real laboratories that do scientifically valid tests on your sample. What's critical is understanding the difference between the testing in the lab, and the analysis that comes afterward. Different companies are going to have different levels of expertise in providing the narrative that helps you interpret the results. If you sent off a swab kit and received back a report that looked like this, it wouldn't be much use... unless you were then willing and able to wade into the vast community of amateur genealogists who analyze these kinds of results (like our Asparagirl) and can translate them for you. That's a significant commitment to learning about genealogy -- some people spend lifetimes on this hobby.

Devonian: "This has been provoked in part by a relative of mine going in for one of the tests, and being very disappointed in the result, or at least as much of it as she understood. I've had a look too, and I can't work out how to check the promises by the companies"

Something else to consider is that the lab work is usually the most expensive part of a DNA test. You can take your relative's lab results and send those off to a different genealogy company for the follow-up analysis, in hopes of receiving better results.

It'd be like having an ankle X-ray from one doctor who says you only have a sprain... except you're pretty sure it's a hairline fracture. The X-ray itself is the same factual snapshot of your body that it was before the first doctor looked at it; therefore you can certainly take that X-ray to another doctor for a second opinion on the same snapshot.
posted by pineapple at 12:32 PM on August 4

Thanks everyone . It rather confirms my suspicions that while there is a great deal of interesting work going on here, and that there is much that this sort of analysis can do, it doesn't necessarily match what people think they're getting when they sign up. I'm also concerned that some companies are perhaps a bit too comfortable with this state of affairs, and a bit of critical consumerist analysis of the market may do a lot of good. And I'm still wondering why it's so hard to find much proper science on the services.

The family tree stuff is one which for various reasons isn't of much interest to me personally, although I understand its fascination, but the larger claims about ancestral membership of sometimes very distantly historical groups are. I'll carry on digging.
posted by Devonian at 4:08 PM on August 5

Just for posterity, wanted to add that my local public radio station is presently airing this re-run of a November 2008 Radiolab show, and it starts out being "about" race but also has a strong emphasis on DNA testing and what it can tell you about your ancestors and countries of origin. (of course they talk about the now defunct DNAPrint, but the material is still interesting if one is learning about genealogical DNA testing)
posted by pineapple at 10:17 AM on August 7