Help us understand unexpected DNA results
October 29, 2019 1:01 PM
A friend took an DNA test and there was an unexpected result. According to the test, their ethnicity is just under 30% “European Jewish”. This was surprising since, as far as they know, they have no Jewish ancestors.
Everything else was pretty much as expected – lots of German and a good mix of other northern European percentages. The brand of the test was Ancestry, which I believe tests autosomal DNA, I have no idea if that makes a difference or not. I’m trying to help this person understand this result more and did a bit of googling, but most articles I can find are focused on either people who know they have Jewish ancestors and are looking to find out more using DNA tests, or people with a result in the single percentages wondering what that means - and the answer for those low amounts seems to be “not much”.
But having a result of closer to 30% seems like a lot. Does that point to a more recent ancestor, maybe one of their grandparents, being ethnically Jewish? Or is this just a weird blip? I know finding out ancestry through DNA is not an exact science.
This question brought to you by the movie School Ties being on TV last night, which we didn’t watch, but made us start talking about this again.
Everything else was pretty much as expected – lots of German and a good mix of other northern European percentages. The brand of the test was Ancestry, which I believe tests autosomal DNA, I have no idea if that makes a difference or not. I’m trying to help this person understand this result more and did a bit of googling, but most articles I can find are focused on either people who know they have Jewish ancestors and are looking to find out more using DNA tests, or people with a result in the single percentages wondering what that means - and the answer for those low amounts seems to be “not much”.
But having a result of closer to 30% seems like a lot. Does that point to a more recent ancestor, maybe one of their grandparents, being ethnically Jewish? Or is this just a weird blip? I know finding out ancestry through DNA is not an exact science.
This question brought to you by the movie School Ties being on TV last night, which we didn’t watch, but made us start talking about this again.
It's not unusual to discover that an ancestor, not even that far off, was Jewish but had to hide it. Think of Madeleine Albright.
Yes genetic markers can't determine but with a giant database of DNA they're making a guess, and they might be right.
posted by wellred at 1:15 PM on October 29, 2019
Yes genetic markers can't determine but with a giant database of DNA they're making a guess, and they might be right.
posted by wellred at 1:15 PM on October 29, 2019
Yeah, I don't think anyone should be surprised that someone in eastern Europe, especially Germany, might not identify as Jewish in the 20th century. Are there living grandparents who could speak to this? Or are your friend's parents still alive? Do they have birth certificates for their parents? Do they know a lot of relatives through their grandparents?
posted by bluedaisy at 1:31 PM on October 29, 2019
posted by bluedaisy at 1:31 PM on October 29, 2019
The Frontiers paper looks like it's more focused on "Jewish identity is distinct from genetics" than discrediting genetic testing results, and the wikipedia info supports that there is a surprising amount of common heritage among Jewish populations.
Since it sounds like your friend is interested, they can pick the most likely grandparent(s) and start digging. They should tread carefully with calling themselves Jewish vs. being interested in their family history, but it sounds like they're on an okay track there.
posted by momus_window at 1:45 PM on October 29, 2019
Since it sounds like your friend is interested, they can pick the most likely grandparent(s) and start digging. They should tread carefully with calling themselves Jewish vs. being interested in their family history, but it sounds like they're on an okay track there.
posted by momus_window at 1:45 PM on October 29, 2019
Depending on where the grandparents came from, there were also thousands of Jewish orphans after World War II and many of them got adopted, sometimes by families that sheltered them during the war, sometimes by others. That might be one possible source of the results.
posted by I claim sanctuary at 1:54 PM on October 29, 2019
posted by I claim sanctuary at 1:54 PM on October 29, 2019
Religious endogamy (as in Judaism) can also "condense" DNA, making those results a higher percentage of someone's DNA make up, because of intermarriage way-way-way back. It's the same with French Canadians, who all descend from a small number of immigrants.
Best bet is for your friend to use Ancestry to actually build out their family tree a few generations back. Then Ancestry's ThruLines feature will show DNA connections to other family trees and potentially clear up the mystery.
posted by xo at 2:04 PM on October 29, 2019
Best bet is for your friend to use Ancestry to actually build out their family tree a few generations back. Then Ancestry's ThruLines feature will show DNA connections to other family trees and potentially clear up the mystery.
posted by xo at 2:04 PM on October 29, 2019
Yeah, one of my cousins was delighted to discover that she's 15 percent African and 85 percent white. She's an artist and has long been drawn to African art, has traveled to various African countries numerous times. She was curious but isn't good at research so she asked me to try to figure it out. We're related through our mothers and I know from doing some genealogy research that the ancestors on that side are white. I discovered that her father's parents were from Jamaica, something she hadn't known. So now we're figuring that one or both of her grandparents were light-skinned people with some African ancestry who decided, as they used to say, "to pass" for white when they came to the US around 1890. (My cousin is 75 and her parents had her when they were in their forties, there's no one left to ask.)
I think it's very possible that your friend has some ancestors who did something similar, they "passed" for Christians, for any number of reasons. Finding relatives might be difficult because there was so much intermarriage among Jews that many show up as related. It's usually something like they're both your fifth cousin twice removed and your third cousin once removed and etc. I'm 1/2 Jewish and those DNA tests give me all kinds of relatives on the Jewish side, none of whom are actually closely related. And names often got changed when they came to the US to something more anglo-sounding, like my neighbor whose great-grandfather somehow chose the name Chapman and my coworker whose ancestor changed Bernstein to Barnes.
posted by mareli at 4:12 PM on October 29, 2019
I think it's very possible that your friend has some ancestors who did something similar, they "passed" for Christians, for any number of reasons. Finding relatives might be difficult because there was so much intermarriage among Jews that many show up as related. It's usually something like they're both your fifth cousin twice removed and your third cousin once removed and etc. I'm 1/2 Jewish and those DNA tests give me all kinds of relatives on the Jewish side, none of whom are actually closely related. And names often got changed when they came to the US to something more anglo-sounding, like my neighbor whose great-grandfather somehow chose the name Chapman and my coworker whose ancestor changed Bernstein to Barnes.
posted by mareli at 4:12 PM on October 29, 2019
In addition to the likelihood of passing (often for social/economic/staying alive reasons), the percentages in commercial DNA tests often don't mean what you think they mean. If you learn that that 30% of your DNA is "Jewish" (which is a weird statement, like celebrating 2.4 nights of Hanukkah), but rather, 30% of your DNA matches with other people on the service who identify as European Jewish ancestry.
(That's also why you get non-base-2 percentages like 15% or 30% instead of the expected 12.5% or 25%. Consanguinity is a thing in many cultures, but it's usually not THAT much of a thing.)
I know this was a big issue with African American individuals, for instance, who sometimes get "updated percentages" over time, as more African Americans joined the service. It's not that an individual person's DNA goes from being 3% African to 40% African (it's the same string of basepairs!); it's that many more people identifying as African American have signed up so the reference dataset is more robust.
23 and Me goes into some detail about this, including what they call "precision" and "recall" -- which I think are sensitivity and specificity? It's a fairly readable guide for those who are interested.
posted by basalganglia at 4:21 PM on October 29, 2019
(That's also why you get non-base-2 percentages like 15% or 30% instead of the expected 12.5% or 25%. Consanguinity is a thing in many cultures, but it's usually not THAT much of a thing.)
I know this was a big issue with African American individuals, for instance, who sometimes get "updated percentages" over time, as more African Americans joined the service. It's not that an individual person's DNA goes from being 3% African to 40% African (it's the same string of basepairs!); it's that many more people identifying as African American have signed up so the reference dataset is more robust.
23 and Me goes into some detail about this, including what they call "precision" and "recall" -- which I think are sensitivity and specificity? It's a fairly readable guide for those who are interested.
posted by basalganglia at 4:21 PM on October 29, 2019
The science behind these genetic analyses is really not very good. They have a very simplified view of genetic ancestry. Also sometimes they are flat out wrong, read 23andMe Made Me Rethink My Identity—Twice for a meaningful example.
posted by Nelson at 4:38 PM on October 29, 2019
posted by Nelson at 4:38 PM on October 29, 2019
"What is my heritage?" is fundamentally not a scientific question.
There are no such things as German genes, or Jewish genes, or European genes, or African genes or anything else. There are only people with a certain ethnic/cultural identity and what genes they have. And closely related ethnically is not necessarily the same thing as closely related genetically. And since human populations are not, and have never been all that isolated, even the most "homogenous" human populations have a lot of genetic diversity. And a lot of these "ethnic identities" are really quite modern formations: Italian, German, all the countries that Africa got arbitrarily divided into by colonial powers, and so on. It's only a couple hundred years, that's nothing in terms of genetics and evolution for a species as slow to reproduce as humans.
Heritage just not something you can define in a neat and tidy way.
So what do the scientists at these companies do when asked a question that's fundamentally not scientific? They come up with a similar question that is strictly defined. In this case, I expect it's something more long the lines of "How well can I predict what a person will put on a form aking their ethnicity from the genetic markers in their DNA sample?"
So, regardless of how they're selling it, don't think of it as "my genes are 30% European Jewish" because that's not a statement that makes any sense. Think of it as "this company's machine learning algorithm predicted that I am 30% likely to put down 'white' and 'Jewish' on a questionnaire about my ethnicity." Because that's closer to the truth.
posted by Zalzidrax at 5:36 PM on October 29, 2019
There are no such things as German genes, or Jewish genes, or European genes, or African genes or anything else. There are only people with a certain ethnic/cultural identity and what genes they have. And closely related ethnically is not necessarily the same thing as closely related genetically. And since human populations are not, and have never been all that isolated, even the most "homogenous" human populations have a lot of genetic diversity. And a lot of these "ethnic identities" are really quite modern formations: Italian, German, all the countries that Africa got arbitrarily divided into by colonial powers, and so on. It's only a couple hundred years, that's nothing in terms of genetics and evolution for a species as slow to reproduce as humans.
Heritage just not something you can define in a neat and tidy way.
So what do the scientists at these companies do when asked a question that's fundamentally not scientific? They come up with a similar question that is strictly defined. In this case, I expect it's something more long the lines of "How well can I predict what a person will put on a form aking their ethnicity from the genetic markers in their DNA sample?"
So, regardless of how they're selling it, don't think of it as "my genes are 30% European Jewish" because that's not a statement that makes any sense. Think of it as "this company's machine learning algorithm predicted that I am 30% likely to put down 'white' and 'Jewish' on a questionnaire about my ethnicity." Because that's closer to the truth.
posted by Zalzidrax at 5:36 PM on October 29, 2019
The way these algorithms work is by comparing a person's DNA to DNA submitted by people from various regions. It's an interesting conversation starter, not meaningful data.
This does not mean that she has Jewish ancestry. It means that she probably shares roughly 30% of the genes they test (no services checks every gene afaik) with some people who submitted their DNA to the service she used and indicated that they were Jewish.
posted by Ahniya at 10:35 PM on October 29, 2019
This does not mean that she has Jewish ancestry. It means that she probably shares roughly 30% of the genes they test (no services checks every gene afaik) with some people who submitted their DNA to the service she used and indicated that they were Jewish.
posted by Ahniya at 10:35 PM on October 29, 2019
In my family, we've had almost the opposite results, in the sense that we are less Jewish than we thought. I haven't sent in my sample yet so I'm not speaking for myself, but generally over time people engage with each other much more than we imagine. I heard an anthropologist on the radio explaining that over the centuries, different groups always integrate. Isn't that nice? We are all humans, and we like having sex with each other.
posted by mumimor at 7:08 AM on October 30, 2019
posted by mumimor at 7:08 AM on October 30, 2019
You have an optimistic view of the nature of the interbreeding, mumimor.
But you're right about how much human populations intermix. I just finished reading Who We Are and How We Got Here: Ancient DNA and the New Science of the Human Past by David Reich and it's fascinating. One of his central conclusions is that human populations intermix way more than we thought; for instance pretty much all ethnic Europeans are descended from a stock of people who only moved in to Europe (from the east) 6000 years ago. Lots and lots of mixing everywhere.
To OP's point the center of the book is describing advances in statistical genetic work on the whole genome. A lot of the analyses of "this is where you come from" come from testing mitochondrial DNA, which comes only from the maternal line, or haplogroups on the Y chromosone that come only from the paternal line. But those two ancestors are only a tiny, tiny fraction of all your DNA. They're easiest to work with since they are mostly unmixed, but they are a vanishingly small fraction of your genetic heritage. Reich's lab has been using statistical methods to get insights into the whole genome and the results they are getting with this larger data set are significantly changing the insight we've gotten so far from mitochondrial and Y chromosome work. Genetic Eve and Genetic Adam aren't the whole story.
Back to OP's question, this whole genome work is all brand new science, from the last few years. I don't believe Ancestry or 23AndMe are using any of it yet. That's why I said above "the science on these things is not very good". That's not quite fair; it's reasonably good, but also now outdated and limited in ways we're only beginning to appreciate.
posted by Nelson at 8:00 AM on October 30, 2019
But you're right about how much human populations intermix. I just finished reading Who We Are and How We Got Here: Ancient DNA and the New Science of the Human Past by David Reich and it's fascinating. One of his central conclusions is that human populations intermix way more than we thought; for instance pretty much all ethnic Europeans are descended from a stock of people who only moved in to Europe (from the east) 6000 years ago. Lots and lots of mixing everywhere.
To OP's point the center of the book is describing advances in statistical genetic work on the whole genome. A lot of the analyses of "this is where you come from" come from testing mitochondrial DNA, which comes only from the maternal line, or haplogroups on the Y chromosone that come only from the paternal line. But those two ancestors are only a tiny, tiny fraction of all your DNA. They're easiest to work with since they are mostly unmixed, but they are a vanishingly small fraction of your genetic heritage. Reich's lab has been using statistical methods to get insights into the whole genome and the results they are getting with this larger data set are significantly changing the insight we've gotten so far from mitochondrial and Y chromosome work. Genetic Eve and Genetic Adam aren't the whole story.
Back to OP's question, this whole genome work is all brand new science, from the last few years. I don't believe Ancestry or 23AndMe are using any of it yet. That's why I said above "the science on these things is not very good". That's not quite fair; it's reasonably good, but also now outdated and limited in ways we're only beginning to appreciate.
posted by Nelson at 8:00 AM on October 30, 2019
This thread is closed to new comments.
Genetic Markers Cannot Determine Jewish Descent
Wikipedia’s article about genetic studies on Jewish people has related background and citations to further reading.
posted by SaltySalticid at 1:10 PM on October 29, 2019