I bought a few “23 and Me” kits for some family members (including teenagers) for Christmas, and coincidentally sat next to a Genetics professor on a flight. She gave me some warnings including 1) taking the results with a grain of salt, 2) being conscious of the privacy risks, and 3) being prepared for what the results might show. Since I had a couple of drinks and wanted to sleep on the flight, I didn’t press her to expand on her concerns. So while I’ve already handed out the kits, I now wonder whether I should also follow up with any of these warnings. I’d love to get opinions from the trusted SDMB community. In case it matters, I purchased the full kits that include ancestry and medical.
My parents both did 23andme and they both came back 99+% Ashkenazi Jewish, which makes perfect sense. My mother then got a follow up email saying that based on her genotype, there are likely no redheads in the family and thin hair is a trait. Both are patently incorrect.
I have a friend who found out about a half sister he had no idea he had through 23andme. So it’s not all junk science.
She is right to some degree on all three points
1 - the salt - possible links between certain things are not positive links; even when it comes to geography. Genetically I am 84% central Siberian even though you have to get back to 1700 to find an ancestor there. And we had other European groups coming into the clan for a thousand years before that. And other Asian groups. So the exact when’s and where’s are more something you find from a paper trail than blood testing. And remember back when women had radical surgeries because there could be a genetic link to breast cancer? Look at the test results and if you find something that concerns you, consult a real-life doctor and not Da Google.
2 - Google rapists and murderers tracked down because some relative had a DNA test. It seems the police and/or FBI have sometimes registered DNA in the hopes of finding, if not a suspect, a close matching relative.
3 - every once in a while someone with a reputation of being called a bastard finds out they really ARE. Or adopted.
A relative is actually famous for initiating the field of using genetics testing to confirm or disprove what is in the written and oral histories; she called it gene-eology. Even hosted a few specials on PBS. And she will be the first one to say that it has some issues. IMHO the good outweighs the bad and I am in the database.
This article discusses some of the pitfalls. Pay especial note of #3.
The thing is, nationality is not genetic. It’s a guess, and the technicians are known to fudge the data. Sending DNA to two different companies will give you two different results.
Absolutely. So much depends both on the quality of the underlying database (you are paying them to use your data to add to it), and on the informational value of whatever taxonomy they choose to use for classifying people in the database. Are they just comparing your data with the geographical distribution of similar DNA at the moment? Do they have enough data on DNA worldwide to provide any meaningful information?
“Nationality” (being even more variable over time than other such categories) is the least useful notion.
What might be more reliable and interesting is when they can tell you something about the movement of ancestors over (very long) timespans.
Ethnicity interpretation is an emerging technology. Your ethnicity results are likely to be broadly correct, but don’t get bogged down too much, especially in the tiny percentages. Keep in mind that, if you have siblings who aren’t your identical twin, each of them would get entirely different percentages and maybe even some different ethnicities. Your own test (or your identical twin’s test) repeated on another day might get different ethnicity results.
The shared centimorgans with other testers is more predictable, but even there the centimorgans can vary between testing companies because of differences in methodology. Having said that, the test can match a parent/child very reliably. Most other relationships fall within predictable ranges. The ranges can encompass several different relationships, such as aunt/uncle-niece/nephew, or half sibling, or grandparent-grandchild, which all fall in the 1300-2300 range.
I don’t have an opinion on the medical side. I am genealogy-focused.
I would be cautious about giving them as gifts. Families have secrets, and they can come out. I’ve started cautioning people who have tested to help me that they may find out things they can’t unlearn. One of my matches is a man who got a kit for Christmas, and by matching with me, discovered his maternal grandfather is not his mother’s biological father. This was a huge deal for his family, as his mother has a complex relationship with her many paternal half siblings - turns out, they aren’t biologically related to her at all, and there’s a chance she’ll be ostracised entirely if they find out.
There have been a couple of threads on this before. At all of them, I said that it didn’t tell me much more than I could find out by looking in a mirror. Their ancestry composition is pretty much junk science for Sub-Saharan Africa.
However, as more people have signed up, there’s more and more of this type of thing happening -
and I’m sure there will be more going forward. And while 23andme is saying that it isn’t sharing data with law enforcement, who knows whether law enforcement is working its way around that issue.
Both my parents have done tests, with different companies. I have not, but I am certain I am my parents’ daughter, so I can expect some of the results.
My father’s results said that he has some Native American blood. He never told his mother, even though she was theoretically not rascist. As she was a geneologist, she had his records quite far back. Assuming the Native American blood is true, one of the more recent (4-5 generations back) may have lied about the parentage.
Which is why I’m not all that thrilled with geneology. It’s too easy to lie about paternal parentage.
Speaking of which, the Mayflower Society has stated that in the future they might require DNA results of applicants. The geneological records will not be enough. Not sure when, though.
Certain parts are remarkably reliable and other parts are junk science. And 23andme doesn’t really make clear which is which.
A couple family members have done it. (One fairly early on.) They connected them to a distant cousin’s DNA (who we already knew about, so a nice confirmation there).
So I’d trust DNA matches up to 3rd cousin or so.
But the probability of traits/genetic conditions is a mess. A few genetic traits are accurate enough but the majority are garbage. One relative was said to be low risk of going bald. Negative on that, buddy. Again, which is reliable and which isn’t is not something they get into.
The ancestry stuff is the worse. I remember an article about three sisters who each got a completely different ancestry report back. (And the test confirmed they were sisters.)
The number of people who have found out that they are at risk for some disease when in fact they might not be is worrisome.
My relatives’ ancestry reports had minor problems.
“Sharing” implies not providing the information after a simple request. I don’t take that to mean they are somehow able to ignore court orders. Working around the issue is getting a search warrant.
I did 23andme hoping to find out I was adopted. Unfortunately, it turns out the people who raised me were my parents.
there is a test for dog breeds that is pretty accurate. Some people get mad when they find out their dog is not 100% pure. 
I know the guys who helped develop it. There are also tests for Cats.
What’s the point? All cats are Bastards! 
I saw something about the cats but just scanned it. 
My HMO was so cheap that when I needed a Cat-scan they had me eat tuna for a week and then used a real cat.
<rimshot>
The ethnicity estimates are in their infancy. They are getting better with time, but I wouldn’t do the test just for that. Like the poster above, I do genealogy and the ethnicity estimates are of little use to me. They’re an amusing curiosity, but not something I take too seriously.
The health stuff might be a little better on some things, not so much on others. Just like with the ethnicity estimates (and without the really obvious mutations like BRCA1 and BRCA2), take them with a grain of salt.
DNA matching to others is extremely reliable.
23andme will also provide your mitochondrial haplogroup and, if you’re male, your Y haplogroup. This, too, is extremely reliable. While it’s not especially useful for tracking recent ancestors, you can follow the evolution of the haplogroups, their geographical locations, and the historical population movements of people in your groups.
I haven’t done it for 2 reasons. I’m concerned about privacy and I’m Ashkenazi Jewish. Seems like everybody I know who thinks they’re pure Ashkenazi turns out to be right - percentages in the mid-90s and above. So why bother?
I’m surprised I haven’t heard more stories of family histories upended because of this stuff. My BFF’s family got a surprise. Fortunately it wasn’t really traumatic, but it did require them to rethink parts if their identity.
they live in Rhode Island. As the story went, a male ancestor traveled out to the west coast and brought back a wife from the Nez Perce tribe. So they all thought they were part Nez Perce. This belief was supported by some authentic Nez Perce family heirlooms.
Upon taking the tests, they discovered that they had no Nez Perce ancestry. However, THEIR genetics were found amongst the present-day Nez Perce! So the ancestor did travel out there and father children with a Nez Perce woman, but didn’t bring any people back with him - just the items that became the heirlooms.
They were proud of being part Nez Perce, so this all was a bit of a disappointment, but not the end of the world. In other circumstances, discovering that you’re not actually Native American could lead to some real problems.
When I took the test, it reaffirmed what my parents had always told me, so I’d say it’s accurate. Certainly not perfect, as some of my results changed (minorly — they decided my 4 percent Italian was actually 2 percent Portuguese and 2 percent undetermined, or something like that).
I suspect the “grain of salt” has more to do with the fact that many people have a poor understanding of probability. 23andme does include some fun results with things like how predisposed you are to having thick hair, perfect pitch, being a night person, stuff like that. And I think many people mistakenly think that if they have curly hair and the report says people with their DNA are “less likely to have curly hair,” then the report is wrong. Let’s say someone with “DNA for curly hair” (no such thing, but using that terminology for simplicity’s sake, has a 75 percent likelihood of having curly hair, whereas someone with “DNA for straight hair” is only 25 percent likely to have curly hair. If you have “the DNA for straight hair” but your hair is curly, all that means is that you were in the 25 percent of people that had curly hair. You WERE less likely to have curly hair, but “less likely” is very different from “impossible.” Having said that, I suspect 23andme has done far more research on the ethnic origins of various DNA strands than they have on how your DNA affects your personality, lifestyle, or physical appearance.
Bottom line, do I recommend the test? Yes. Particularly if I got it as a free gift. Also, 23andme does have this brief tutorial you have to take before they show you your results, so that you’re informed of the risks and can back out if you want to. If everyone is a teenager or older, then I think this is sufficient. As long as no one who receives this gift is mentally impaired, they can read the information given and come to their own conclusions about what information they want to receive.
Great info, everyone. Thank you!
BTW, decided to take the one I gave my daughter back and replace it with another gift. She’s 13, and was already nervous about it. I’m going to do it myself, and let her decide for herself when she’s ready.