As of 2008, what can science tell me about my genetics thru DNA testing.

April 16, 2008 6:20 AM

Wondering if anyone here has had a Genealogical DNA test for purposes of learning more about your distant (ie. not recent/paternity) ancestors. I have questions about the differences (price, service, thoroughness) of the various companies offering this service, the type of data that different tests can produce, and ways this data can be (constructively) interpreted.

I had heard of various folks having these tests done (one example of many) for all kinds of reasons. I was/am interested, and I researched it a bit. But I got overwhelmed somewhere between Y-chromosome DNA haplogroups and Mitochondrial DNA haplogroups, and started wondering just what, exactly, these tests could actually deliver.

Then, to confuse me even further, at minute 1:10 of this YouTube video, Christopher Hitchens says the following, which, if true, raises my estimation (and expectations) for these tests a hundred-fold:

"...and I, by the way, have been to the National Geographic... and had--as you can have, for a hundred bucks--my DNA analyzed. And I can show you on a map which part of Africa my ancestors come from. Looks like Angola, In my case."

So, MetaFilter... please clear my head. My questions (to start) are these:
• How much (and how specifically) can these tests tell me about my ancestry (best case/worst case)?
• What types of tests exist, and what are the strengths and weaknesses of each type?
• Which suppliers/labs dealing in these tests do you have a personal (even anecdotal) experience with that has led you to have either a high, or a low opinion of them?
• What price range are these tests?
• What combination of test_type/lab/other_variables will most likely produce the largest, most enormous and rich mountain of (worthwhile) interpretable data?
• What are the things that I am certainly overlooking that I should not be?

If this helps at all: I'm a white guy, mostly Northern European ancestry. I'm totally ignorant about, and incredibly stoked by the "non-mostly" part of that equation, and I'm chomping at the bit to get this done. Yes, it's vanity, but please indulge me this one thing. I'll wear the same shirt all next week in penance.

While I'm not very well-versed in the biological sciences, I'm not afraid of reading links you send or learning about it. In other words... not a "tell me like I'm 6-years-old" type of deal--more like a "tell me like I am an conscientious, interested, incoming freshman to your "intro to bio" class"... (or somewhere in-between). Many thanks.

posted by cadastral to science & nature (17 comments total)

Keep in mind that any of these tests only show one of your ancestral lines of which you have multitudes.

The mitochondrial DNA test for example will only show your mother and your mother's mother and your mother's mother mother etc.

So, as a for example, lets take a look at your 8 great-grandparents. The paternal line tests will only reveal (part of) the ancestry of one of those 8. The maternal tests will reveal the ancestry of the 2nd of those 8. So you've only covered 1/4 of your great grandparents.

By the same reasoning, you only cover 2/16=1/8 of your great-great grandparents. Did your Dad's Mom's Dad's Mom come from some really exotic location? It doesnt matter. She's invisible according to these tests.
posted by vacapinta at 6:47 AM on April 16

A cousin had it done. It told us basically nothing, except that our maternal line had done time in Norway a few thousand years ago. It was so vague as to be useless. I would wait until something more specific comes along.
posted by clarkstonian at 7:04 AM on April 16

(How did C. Hitchens... arrive at that "Back to [a specific country in] Africa" bit? Was he just blowing smoke [as he is wont to do]?)
posted by cadastral at 7:12 AM on April 16

You may have come across it, but somewhere I read a review of several of these services where they sent the same sample to a bunch of the different labs and received a bunch of different answers. For the life of me I can't remember where I came across it, but I'm not surprised.

I just finished a masters in genetics, and I studied exactly this kind of technology. With the currently available technology at the costs that they are talking about, there is no way they are looking at your whole genome. More than likely they just have a few bits of sequence that they have decided originated in a particular region of the world, and they check yours for those bits of sequence. I'd almost go as far as to say it's a novelty at this point... kind of like those services that can name a star after you.

The $100 genome is coming up really quickly, though. There are already companies claiming that they will soon have the technology to sequence the entire human genome in a few minutes for under $1000. Researchers are clamoring for a centralized database of genomes combined with disease histories and demographic information. If that happens, then a LOT of genealogical information can be extracted from your genome that is probably very accurate. The biggest obstacle to that happening is the legal/ethical implications. I'd be able to pluck a hair off your head, sequence it, find that sequence in the database, and then I'd know your medical history.

So to answer your question:

How much (and how specifically) can these tests tell me about my ancestry (best case/worst case)?

I'm sure they will give you a fascinating packet of material. I doubt they will give you the statistical information to tell you how confident they are that their guesses are true. When working with the genome, it's really easy to say, "this matches, this matches, and this matches." The hard part is determining whether they match my random chance or whether it's significant.

What types of tests exist, and what are the strengths and weaknesses of each type?

At the prices they are talking about, I imagine they are doing PCR using a few bits of sequence that have been traced to different regions as their primers. You can then step up to a SNP array, which checks for the presence of a bunch of single nucleotide polymorphisms, which can be used as genetic markers for disease and ancestry (among other things). If you want to get more expensive (hundreds of dollars), you can go with a microarray, which can check for thousands of bits of sequence at once. If you want to get even more expensive (thousands, of dollars), you can go with a tiling array, which is a series of microarrays that cover the entire non-repeat human genome in overlapping segments. Finally, if you want to get ridiculously expensive, you can have your entire genome sequenced and then work with that information.

As you know more of your sequence and what level the genes within it are expressed, there are several statistical tools out there that can be used to compare this information to what we already know about other people. It'll spit out probabilities of how similar your profile is to other profiles, and then we can use that information to deduce things about you. The strengths of this approach are that the math is all sound, and we are fairly certain it works. The weakness is that we really don't have much information to compare it to yet, because gathering all of it is very expensive.

Assuming these firms are going with the PCR approach, they are only able to check for a handful of sequences. That'd be very misleading. A very simplified analogy would be if you were to invent a "food identifier". You stick some food in it, and it tells you what it is. I could put in a rule like, "if it has bread, cheese, and pepperoni, then it's pizza". There are a lot of things that you could stick into that machine that would come back flagged as pizza that are clearly not pizza. In the same way, the human genome is too complex to be able to draw conclusions just by checking for a few bits of sequence.

Which suppliers/labs dealing in these tests do you have a personal (even anecdotal) experience with that has led you to have either a high, or a low opinion of them?

What price range are these tests?

Can't help you here.

What combination of test_type/lab/other_variables will most likely produce the largest, most enormous and rich mountain of (worthwhile) interpretable data?

Right now, you can spend a lot of money and get not much accurate data out of it. In a few years, that will certainly change. You'll know as it happens, as the mainstream media will definitely take note and I'm sure it will spawn a great deal of MeFi posts. The controversy surrounding it will dwarf the stem-cell debate.

What are the things that I am certainly overlooking that I should not be?

I guess the only thing I'd make sure to take note of is that these companies have no oversight. It's just a way for them to make a buck. One resource that you may find handy is PubMed. It's an index of every scientific journal out there. Just search for combinations of keywords like "genome", "ancestry", and "snp" and you may find an article that discusses exactly what you want to know.

I really did not intend for my reply to be so long.
posted by AaRdVarK at 7:15 AM on April 16 [9 favorites]

I have read a smorgasbord of complaints about one of the services... I can't remember which one it was but it was a highly prominent one. Try putting in a Google search for (name_of_company) sucks and it should put you on the right track to find people who are dissatisfied.
posted by crapmatic at 7:17 AM on April 16

I've used Family Tree DNA for this service. I found it informative and get high resolution DNA matches from others quite frequently. This allows you to contact folks you may be related to distantly.
Tests are easier for guys, you can get a discounted rate if you agree to join the study group for your surname.
posted by nougat at 7:47 AM on April 16

Boy uses internet to find his sperm-donor father

He took a saliva swab and sent it to FamilyTreeDNA.com, an online DNA-testing service. His genetic father had never supplied his DNA to the site.

But after nine months, the boy, who was unnamed in the report, was contacted by two men on the database. Their Y chromosomes were a close match to his, suggesting a 50 per cent chance that all three had the same father, grandfather or great-grandfather.

posted by bleary at 7:55 AM on April 16

AaRdVarK : I wasn't sure exactly what kind of guidance I'd be getting from this thread... but I gotta say, your answer was far more thorough and informative than any answer I had even considered the possibility of getting. Thank you so much for sharing your expertise, and for the time it surely took you to compose such a great reply. I owe you one.
posted by cadastral at 7:58 AM on April 16

So, as a for example, lets take a look at your 8 great-grandparents. The paternal line tests will only reveal (part of) the ancestry of one of those 8. The maternal tests will reveal the ancestry of the 2nd of those 8. So you've only covered 1/4 of your great grandparents.

By the same reasoning, you only cover 2/16=1/8 of your great-great grandparents. Did your Dad's Mom's Dad's Mom come from some really exotic location? It doesnt matter. She's invisible according to these tests.

@vacapinta: I don't understand how you're coming by these figures. Before gametes (eggs and sperm) are produced by the process of meiosis crossing over occurs between your maternal and paternal chromosomes. This exchanges bits of DNA between the chromosomes. This occurs for all 23 pairs of your chromosomes. Thus all the eggs/sperm that are produced have bits from both your maternal and paternal ancestors. Therefore it is incorrect to say that a maternal test will only tell you about the ancestry of 1/4th of your great grandparents or 1/8 th of your great grandparents. It will tell you about 1/4 th of each of your great grandparents' DNA or 1/8th of each of your great-great grandparents' DNA. But none of your ancestors becomes invisible as such. Though things get muddled up so quickly it doesn't seem like a very worthwhile proposition.
posted by peacheater at 8:22 AM on April 16

I read a review of several of these services where they sent the same sample to a bunch of the different labs and received a bunch of different answers.

Next step: send the same sample to the same labs under different names. I bet that would be entertaining. I've tried similar experiments in hospitals (drawn two samples out of the same needle, then labeled them with different times and sent them in for assay) with amusing results.
posted by ikkyu2 at 9:35 AM on April 16

Kudos to AaRdVarK for their answer. I'll underline that answer and say that these commercial tests are mostly phlogiston. Even a little simple maths will show you how they over-simplify: go back 10 generations, which might put you in the 1700s. You have 2 parents, who had 2 parents and so on to give you (assuming no inbreeding) 1000 ancestors in the 18th century. Saying you "came from Angola" is a gross simplification of the movements of those thousand people. Not mentioning the extremely low sampling used in most of these tests. In reality, the answer probably means something more like "1% of your ancestors have the same gene as other people that we think lived in Angola. Dunno what the other 99% were doing." It's like plucking a drink can out of the Thames and declaring the river came from the Coca Cola factory.

There's also the question of "when". In the long run, all of us "came from" the Rift Valley.

(Which is not to say that we can't trace the genetics of ethnic groups or association of genes with given peoples over time - but it's not as trivial as it seems, and comes with a whole heap of assumptions and qualifications.)
posted by outlier at 10:03 AM on April 16

National Geographic's Genographic Project website is also a great resource. Particularly this image which helps explain what vacapinta and peacheater are talking about.
posted by jenne at 10:13 AM on April 16

peacheater:

@vacapinta: I don't understand how you're coming by these figures. Before gametes (eggs and sperm) are produced by the process of meiosis crossing over occurs between your maternal and paternal chromosomes. This exchanges bits of DNA between the chromosomes. This occurs for all 23 pairs of your chromosomes. Thus all the eggs/sperm that are produced have bits from both your maternal and paternal ancestors. Therefore it is incorrect to say that a maternal test will only tell you about the ancestry of 1/4th of your great grandparents or 1/8 th of your great grandparents. It will tell you about 1/4 th of each of your great grandparents' DNA or 1/8th of each of your great-great grandparents' DNA. But none of your ancestors becomes invisible as such. Though things get muddled up so quickly it doesn't seem like a very worthwhile proposition.

No.

You're talking about the DNA in the nucleus of your cells. That's not all of the DNA that we have.

We also have DNA in the mitochondia of our cells. This is completely independent from the nuclear DNA, and it comes completely from your mother.

Your mitochondrial DNA is exactly the same as that of your mother, unless mutations occurred. Thus it's also the same (ignoring mutations) as that of your mother's mother, your mother's mother's mother, your mother's mother's mother's mother, and so on, back indefinitely.

As for the paternal side, it is done with nuclear DNA, but with a special case: the Y chromosome. The large majority of the Y chromosome (from your father) does no recombination with the matching X chromosome (from your mother). Thus, your Y chromosome (assuming you are a man and therefore have one) is, in a large part, exactly the same as that of your father, and his father, and his father, and his father, and so on.

Thus, these genetic tests done by National Geographic and so forth can tell you virtually exact information about your mother's mother's mother, and (if you are male) your father's father's father. But they tell you nothing at all about your other six great-grandparents.
posted by Flunkie at 10:39 AM on April 16 [1 favorite]

You may have come across it, but somewhere I read a review of several of these services where they sent the same sample to a bunch of the different labs and received a bunch of different answers. For the life of me I can't remember where I came across it, but I'm not surprised.

Was it this episode of 60 Minutes?
posted by mr_roboto at 10:42 AM on April 16

Was it this episode of 60 Minutes?

Yes! I have no idea what I was doing watching 60 Minutes, but I just watched that segment, and I've definitely seen it. I agree 100% with the guy on there they had who was skeptical of these companies' claims.
posted by AaRdVarK at 1:15 PM on April 16

The cbc had a show about this recently
posted by canoehead at 2:10 PM on April 16

I come from a typical Southerner's Scotch-Irish background, and it's said in my dad's family that one of his great xN grandmothers was passing for white. We're interested in confirming this, but I'm wondering, from this thread, whether, unless --

a. Great-Grandmaw Townsend was of direct, unbroken African descent in the maternal line, and there is a direct maternal-line descendant of hers in the family that we can test

b. Great-Grandmaw Townsend was of direct, unbroken African descent in the paternal line, and there is a direct paternal-line descendant of hers in the family that we can test -- not likely, since she was a woman! --

-- we have no way of finding out whether it is true that she was partly of African descent.

Sure, it quickly becomes statistically meaningless, but I'm interested in "recreational genetics" as a way of connecting with particular ancestors and ancestries. I took the mitochondrial DNA test (with Family Tree) to see if the "Cherokee chief's daughter" story (you know, the one all us crackers have) was true. It isn't, at least in the maternal line. The maternal line goes straight back to "Ulrike," Haplogroup U4, which is eastern/northern European. I find that fascinating enough. It allows me to make a connection with a single Paleolithic human.
posted by Countess Elena at 4:35 PM on April 16