Twins get 'mystifying' DNA ancestry test results

Question

Interesting piece - I was quite surprised to see the results.  How can identical twins have differing DNA test results?

https://www.youtube.com/watch?v=Isa5c1p6aC0&feature=youtu.be

Comments

Just a side comment, I am surprised by all the nay-sayers on this thread.  It is good to have a healthy skepticism, but all DNA tests and resulting traditional research are not created equally.  Believe me when I say that yDNA testing is the ONLY way we have begun to sort through the many unconnected and misconnected paternal Smith lines.  I am sure the same can be said for any other common surname.  Please get your Smith men yDNA tested to further our research.  

---

Agreed Kitty!

---

I dread seeing these sorts of posts (and how many times has it been posted?) because people don't understand the difference between ethnicity testing (still more of an art than a science, although growing ever more accurate almost by the month) and testing for DNA matches between close relatives. Especially in genealogy circles, when people get matches they didn't 'want' to get, they try to handwave it away. You don't match your paternal first cousin? Tell yourself it's just a glitch, because it's better than believing there's an NPE somewhere in your family tree. YDNA doesn't match the royal family from which your great-great-granny swore you descend? Eh, that DNA testing isn't REAL anyway. Found a half-sibling who's existence is inconvenient to your narrative that your parents were happily married? Must be a mistake, this person isn't related to me, my dad would never...!

So when they see 'twins take DNA test and results don't match!' posted for the umpteenth time, these people walk away with these beliefs bolstered. Because someone's Italian DNA is marked as 'Southern European', one convinces oneself that this DNA testing stuff is all malarkey and even though 23andMe says you and your dad share ZERO DNA, well, that's just nonsense because your family tree goes back to Moses. If you scoff at this, please keep in me I've seen someone earnestly arguing here on G2G that someone might inherit no Native American DNA from a supposedly Native grandfather, because males don't pass on their mtDNA so of course the Native DNA doesn't get passed down... forgetting all the autosomes and sex chromosomes we all inherited from our grandfathers.

Answer 1

Doug McCallum, I'm with ya. Other reputable genealogical DNA authorities include Jim Bartlett, Blaine Bettinger, Kitty Cooper, Debbie Kennett, Leah Larkin, and many more. 

I shudder. We have many new millions engaging in DNA testing for the first time, and advertising companies, of all sources, have led them to believe that the 0.023% sample provided of their overall DNA will result in explicit, unequivocal results explaining their ethnicity and heritage.

It won't. Within the current constraints of the technology, it can't. It can tell you if you posses Neanderthal DNA, which if you're of European descent you almost certainly do, but it can't tell you definitively if you're Greek or Italian, or much less from County Cork, Ireland or Kirkcudbrightshire, Scotland. Consider that the Norse circa 1,100 years ago ranged from Iceland to Ireland to the Straits of Gilbraltar to the Mediterranean. I can't count the number of people absolutely shocked to find "Scandinavian" in their DNA.

Contemporary evidence is showing that we, as a species, moved about far more than was believed even just a decade ago. And we moved about far earlier than previously thought. There were no simple and discrete and tidy migrations out of Africa, and then from the Caucasus into Europe or the Levant into East Asia. Didn't happen. Humans are not the rational, organized beings we might believe them to be: ancient human migrations were sporadic, random, and opportunistic, not logical, planned, and structured.

And the video? "Identical" twins is subjective; there is no evidence these were monozygotic twins. The stipulation for genetic similarity...and even then, it's known that monozygotic twins can still differ in copy number variants. My general opinion on ethnicity results here: https://casestone.com/threlkeld/blog/93-should-you-trade-your-lederhosen-for-a-kilt.

And there are many who feel they fully understand genetics and genealogy: https://casestone.com/threlkeld/blog/112-genetic-genealogy-and-the-dunning-kruger-effect. I'll posit only a small fraction of those who think they understand DNA actually do. I confess that I certainly do not, but I've offered on occasion to give an extemporaneous, real-time, oral quiz to some self-proclaimed "experts," and never have been taken up on that offer.

Leave the ethnicity and admixture to the PhDs who study population genetics for a living. Read David Reich's 2018 book. Studies of deep ancestry and anthropology have little or nothing to do with genealogy.

Our common, cheap, direct-to-consumer tests are highly reliable in doing what they do: identifying an allele associated with a specific locus on a chromosome, a SNP. That's what the tests do. Take a squishy, tiny, biological marker and convert it to a digital element that we can read and compare.

The microarray genotyping process doesn't have anything to do with ethnicity or heritage. That's what the marketing-driven testing companies struggle to do with the data after the test is completed. When it's all about what-ifs and comparisons against whatever particular database they're using on the day.

I love that AncestryDNA's marketing campaigns have brought many millions to DNA testing. I hate that the same marketing has given them false impressions about the purpose and usefulness of those tests.

Let's please not confuse the science of DNA testing with ethnicity smoke-and-mirrors guesswork.

Comments

Edison, the class I took was coordinated by Karen Stanbary but Blaine and a number of other reputable genealogists who do a lot with DNA gave lectures. In a panel discussion with 5 people who know what they are talking about, all agreed that ethnicity estimates were “fun” information but pretty much useless to a geneologist. Thank tou for adding some more technical perspective. People do seem to equate DNA tests with the ethnicity estimates and lose the actual scientifically valid and genealogically useful part in the noise.

---

Ethnicity estimates = marketing hype and are a good example of science being high jacked and fiction being celebrated

DNA matches are much more real science when it is done correctly and data down to chromosome and SNP measurements are included but not all labs do that so there is potential for it to also be abused and marketed to an unaware public.

Great answer Edison.

---

The twins are identical, but there were some differences in the no-calls and miscalls. I don't see a way to insert the image, but the video interview with the professor shows the details.

https://www.cbc.ca/news/technology/dna-ancestry-kits-twins-marketplace-1.4980976

23andMe tells me these are enough to impact their methodology. AncestryDNA uses a random technique to get confidence intervals, selecting different subsets of SNPs.

---

"And the video? "Identical" twins is subjective; there is no evidence these were monozygotic twins."

Wrong! Identical/monozygotic twins is not subjective. You are either monozygotic twins or you are not. Barring rare outliers like substantial chimerism, twins are essentially either monozygotic clones, or dizogytic full siblings, which is objective, obvious from testing, and identical twins are generally strikingly identical in appearance. Test results of >98% concordance is not evidence that they are monozygotic, identical twins?

"it's known that monozygotic twins can still differ in copy number variants."

And the SNPs tested by microarrays are generally not located near or affected by copy number variants in STRs. The vast majority of variation between either identical twin or individual retests on the same microarray should be due to no-calls or miscalls.

What this video is revealing is that the admixture analysis of some of these firms is not only unsound but unstable.

---

You're correct, Nathan: there is no genetic subjectivity regarding monozygotic twins, but there certainly is genealogical subjectivity. Witness the not-uncommon instances of families claiming twins of the same sex to be identical when, in fact, they have no evidence of the fact. In that way, in genealogy up until the past couple of decades, yes: whether or not twins are identical has been routinely subjective.

It may also be worth noting that today's genotyping microarrays provide no distinctions between the diploid chromosomes. In other words, the same results in terms of shared percentages and centiMorgans will be returned whether the individuals' alleles are half-identical or fully-identical; parent/child relationships will show the same results as those between monozygotic twins.

Monozygotic twins form from the same ovum. They carry the same genome at embryogenesis. Monozygotic twins almost always possess different phenotypes though, as you note, only rarely different genotypes. Beyond the no-calls and miscalls that Dr. Turner described--the artifacts of microarray genotyping--and the oddity of chimerism ("oddity" discounting red blood cell genomic differences following allogeneic hematopoietic stem cell transplantation), there still exist mechanisms by which monozygotic twins might have differing genomes. Additional reading: "Why are monozygotic twins different?" S. Silva, Y. Martins, A. Matias, I. Blickstein. Journal of Perinatal Medicine. 2011 Mar; 39(2):195-202; doi: 10.1515/JPM.2010.140. "Finding the needle in the haystack: Differentiating 'identical' twins in paternity testing and forensics by ultra-deep next generation sequencing." J. Weber-Lehmann, E. Schilling, et al. Forensic Science International: Genetics. 2014 Mar; 9:42-46; doi: 10.1016/j.fsigen.2013.10.015.

"And the SNPs tested by microarrays are generally not located near or affected by copy number variants in STRs."

I actually wouldn't know how to corroborate or refute that statement without a boatload of research. Do you? I'm unaware of any published studies that examine the matter; would be interesting.

Taking into account the average SNP overlap between Illumina's OmniExpress and GSA default configurations, today we're testing a total of about 1.1 million unique SNPs out of what the NIH says is an individual per-genome pool of 4 to 5 million, of over 100 million globally identified. There are only 30 or fewer autosomal STRs commonly tested for forensic purposes, but there are roughly 700,000 STRs identified ("The landscape of human STR variation." T. Willems, M. Gymrek. et al. Genome Research. 2014 Nov; 24(11): 1894-1904; doi: 10.1101/gr.177774.114.). In dbSNP I can readily find 2,072 STRs identified by locus reference sequence, but evidently there are 5,500. Many of the SNPs selected for population genetics and genealogy are chosen because of their known, relatively high degree of polymorphism, and STRs are among the most polymorphic of loci along chromosomes. So it would not surprise me at all to find that, in fact, many identified STRs are at or near SNPs examined by our common genotyping array tests. That said, variances in repeat numbers shouldn't affect the call at any single SNP locus. But since we're nitpicking selected quotations...

In any event, it seems we reached the same final conclusion and opinion, Nathan. Usefulness of current, direct-to-consumer, inexpensive autosomal DNA tests for genealogy? Highly useful and suitably accurate (given that we still wrestle with the SNP overlap disconnect between OmniExpress and GSA tests).

Usefulness of the same tests for ethnicity and heritage information? Shaky; as much marketing and guesswork as science; dependent upon different, proprietary, undisclosed databases; and probably best viewed as "for entertainment purposes only" at this time.

---

"It may also be worth noting that today's genotyping microarrays provide no distinctions between the diploid chromosomes. In other words, the same results in terms of shared percentages and centiMorgans will be returned whether the individuals' alleles are half-identical or fully-identical; parent/child relationships will show the same results as those between monozygotic twins."

Full siblings or fraternal twins are trivially distinguishable from identical twins and it is statistically impossible for them to be confused. The law of large numbers ensures that fraternal twins/full siblings have close to 50% match while identical twins will match 100%.

It is true that depending on how you count half-identical regions, both parent/child and siblings may be listed as 50% matches, however, they are trivially distinguishable from raw results (such as a chromosome map) because a parent/child is always half-identical across the entire genome whereas siblings have a random assortment of matching and non-matching segments.

Your two citations on differences in identical twins may be interesting but are not relevant to current genetic genealogy.

"I actually wouldn't know how to corroborate or refute that statement without a boatload of research. Do you?"

I don't have space to get into it here but it fundamentally has to do with how microarrays and primers for SNP testing work. A copy number variant will not affect the sequence that the primers are keyed to. It would normally take a deletion or mutation in the context of the SNP.

At the end of the day the fact that identical twins do match to 99.7% with commercial microarrays is proof enough that any real genetic variation is small and irrelevant for autosomal genetic genealogy techniques; whether 0.1% or 0.0001% of this variation is due to real differences versus testing artifacts does not change genealogical conclusions.

If you really wanted to get into the nitty gritty, we have a pretty good idea of individual mutation rates on nonrecombinant Y due to NGS projects on Y DNA, but that's a different subject.

---

I give up. I'm talking into a vacuum here. You say the SNPs examined in microarray testing are not near STRs. I say I'm not so sure, and that it would be difficult to corroborate, but I note that it wouldn't affect allele values returned by the microarray test anyway. You respond that you don't have enough space here to educate me, and you note that it wouldn't affect allele values returned by the microarray test anyway.

"If you really wanted to get into the nitty gritty, we have a pretty good idea of individual mutation rates on nonrecombinant Y due to NGS projects on Y DNA, but that's a different subject."

Wrong! (I'm really not that passionate about it; I just wanted to return the friendly favor from uptopic.) I'm right there with ya and run four FTDNA projects and co-admin a fifth. Just exchanged email yesterday with Alex Williamson of the Big Tree. We have only an extremely broad and generalized notion of Y-SNP mutation rates. There has yet to be any peer-reviewed study of mutation rates regarding the new and novel SNPs revealed in the course of the past few years of NGS yDNA testing; everything to date is experiential summary. For that matter, I follow most of the online genetic academic journals hoping for better refinement of what evidence we have behind Y-STR mutation rates. There has yet to be anything published in that arena within the past several years, either.

Besides, your references to haploidy, non-recombinant Y-chromosomal SNP mutation "may be interesting but are not relevant to current [autosomal] genetic genealogy."

But thanks for your down-vote of this answer anyway, Nathan.   And have a great day!

Answer 2

I'm glad to see it called 'recreational', 'entertainment' and 'fun'.  Far too many people are practically worshipping DNA results.

Comments

I noticed that too Ros, I don't trust it myself.

Answer 3

It exposes the lack of reliability of DNA testing, the insufficient size of data bases and the difference of testing methodology between the different companies. At best DNA testing is a very small part of researching your family history. The more difficult work of following the paper trail has to be done. There is not an easy way to success.

Answer 4

They used to have a place for people who sold a product they didn't have

https://en.wikipedia.org/wiki/George_C._Parker

But now they just call it Marketing.  You do DNA to become one of the people who've done DNA, so you don't feel left out.

Comments

I've "done DNA".  I tested with Ancestry and Living DNA.  They both told me I came from Devon and Cornwall, with a bit of London in there.  (My father was Cornish, my mother from Devon, I was born in London.)  My paper trail of over 40 years had already told me that...

---

The technology must be way more advanced than I thought.

---

However, they redeemed themselves ha ha by telling me that my great grandmother came from Africa.  The same great grandmother who had ancestors going way-back-when in Cornwall, was born in Cornwall, married in Cornwall, had children in Cornwall, died and was buried in Cornwall. *facepalm*

---

>had ancestors going way-back-when in Cornwall, was born in Cornwall, married in Cornwall, had children in Cornwall, died and was buried in Cornwall.

"Yep.  Africa."

---

Amen and hallelujah. I'd be laughing hugely and screaming loudly all at the same time on reading something like that on my ancestry 

---

you are lucky to have a great grandmother let alone one willing to do a dna test :)

---

No, she wasn't living, and hadn't taken a DNA test.  This was from the test I took.  Ancestry told me she was born etc in Africa.

---

How did you narrow the (supposed) African dna down to that particular great grandmother? I'm fairly new to dna tests, had mine done about a year ago but have not explored it in depths like this.

---

I didn't, it was Ancestry being weird.  They showed me an oh-so-beautful map of where my dna came from, and there she was, right in the middle of Africa.  The one who never left her village in Cornwall. I'm not talking our ancestors from thousands of years ago, either.  Her dates are 1837-1915...

Oh, Ancestry...

---

Oh, I think I know what you mean now, it's that map. It doesn't seem to be tied into the dna, it seems to follow place names instead. For example, because my dad died in Canada, my half sister shows as having Canadian roots, even though both of them were born in Britain. I don't think the algorithm is using the dna at all but is trying to interpret names of places in your tree, sometimes inaccurately.

Mine used to show all kinds of mapping all over the world for people I know never left England, though when I look today, that feature seems to be (thankfully) gone, there's no more "migration" type trails taking people where they never set foot. I wonder if it was a buggy feature they've since discontinued?

Answer 5

It's also helpful to remember that:

• Identical twins might not actually be identical - Most of the time its assumed based on appearance, but you could just have fraternal twins who look very similar like other siblings, and they only find out when it is tested.
   
• The impact of the 'environment' on gene expression
   
• And genetic differences which come about in the process when the cells split
   
• And identical is just the common term for monozygotic twins where the twins develop from one zygote 

Comments

I just wanted to say the same as you mentioned behind the first bullet. Maybe they just aren't identical twins

---

I have twin sons. Lots of people think they look "identical" but my wife and I do not. We think they're just have a strong brotherly resemblance. Also, they were in separate sacs, so that makes their being monozygotic unlikely (though not impossible). So we're pretty sure they're fraternal twins - but friends and extended family struggle to tell them apart.and often ask us if we're sure they're not identical.

---

Thomas Fuller,

We had kind of the opposite situation. My twin and I do NOT look alike, and participated in a university study of twins for most of our youth. That study ultimately assured us that we were identical twins, despite NOT looking alike (ironically, I've been mistaken for my significantly-older brother many times).

AncestryDNA identified us as simply brothers, and our shared DNA isn't much different than between me and my older brother - a normal level for full siblings.

So it turns out that just because a scientist at a university says something, that doesn't always make it true.

---

• Identical twins might not actually be identical - Most of the time its assumed based on appearance, but you could just have fraternal twins who look very similar like other siblings, and they only find out when it is tested. 
   
  • This is true, but the twins here were. The test results definitively confirmed this.
     
• The impact of the 'environment' on gene expression 
   
  • Are you referring to epigenetics? While tests for, e.g. methylation, in a given DNA sample exist, epigenetics and gene expression is irrelevant to the heritable DNA SNPs tested in microarrays by firms used for admixture analysis.
    
    Environment may have important effects on phenotype including health, and may be even affect mutation rate in the germ-line DNA, however it is largely irrelevant to genetic genealogy and admixture analysis.
     
• And genetic differences which come about in the process when the cells split 
   
  • I'm not sure what you are driving at. By "genetic differences" do you mean "mutations"? Yes, mutations happen, that's how variation arises. But within one generation the number of mutations will be small.
     
• And identical is just the common term for monozygotic twins where the twins develop from one zygote 
   
  • Yes.

Answer 6

I am not a believer in DNA testing,  glad to see it is called "recreational, entertainment and fun".

Comments

I gave you an upvote so I see somebody disagrees. I don't trust it on many levels because, is there a chance they could create a clone of you in a secret lab somewhere? Or have I seen Jurassic Park too many times.

---

Oh well - thank you for the up vote.  It is just a personal opinion and cloning has crossed my mind too.  And perhaps I should have added that DNA for health issues is remarkable. One example, my cousin had a DNA test to see if she had Cadasil, as it ran in one of the family lines.  She was given the "all Clear''.  Can you imagine the relief she felt to know she would not be afflicted with this disease?

Answer 7

The advertising for DNA testing IS misleading, but so is this video!

As A. C. Raper points out in their answer, they didn't give us a meaningful statistic that even tells us that they're really identical twins! The subject of the twins' DNA match to EACH OTHER wasn't even brought up.

But maybe they are anyway - perhaps it's just that the measurement of the DNA isn't 100% perfect. The ethnicity breakdown for the two twins - with the SAME company - isn't all that different.

The emphasis is on how the admixture varies. Even here they all pretty much said Sicily/Italy, Greece/Balkans, and Eastern Europe. The actual fractions varied quite a bit (not surprising) but most of them indicated the same general places, which they seemed to be saying were correct.

So, YES, the percentages are inaccurate (and probably always will be, but that's another discussion), and the geographical areas are somewhat vague. It really IS fair to call THAT PART a strictly "for entertainment only" thing.

BUT!!! … DNA tests give you more besides, and THAT is where the real genealogical value lies. The most important thing - by far - is the DNA matches to other individuals, and they don't even MENTION  that!

You don't do a DNA confirmation with your ethnicity percentages - you do it with your DNA matches.

Notice also that in the AncestryDNA test, they pegged that they had ancestry from Sicily. AncestryDNA doesn't make it obvious, but that result actually has nothing to do with the calculation of the percentages. When they give you a "region" like that - with no associated percentage - my experience is that they are surprisingly accurate.

Of course, this is the CBC, so they had to thrown in a little Globalist message at the end, too. Really, this clip is just the usual anti-genealogy propaganda we're seeing from time to time, which the misleading advertising of the testing companies leave us open to.

Comments

Frank, I think you didn't watch the whole video. About 60% of the way in, they discuss that the ladies are identical and share the almost exact DNA.

---

JT, what I remember about it was that they gave various statistics that gave the impression that their DNA is virtually identical. But I don't know what those statistics mean. If they told us what their match is in centimorgans, we'd know. The fact that they didn't might suggest that we're being misled, or it might just mean that they consider that to be too technical for their audience.

Really, their admixtures are a lot closer than what I've seen for full siblings, so that makes me suspect that maybe they really ARE identical twins. But the whole "twin angle" of the video is superfluous for the point they're trying to make - it's just kind of a gimmick, and just confuses things. You could make the same point with just a single person doing tests at different companies.

---

Good on ya, Frank. I agree with everything you've said.

---

At 13:55 they give a table with the exact numbers of concordant and discordant SNPs for each testing company. The lowest is 98.4% (LivingDAN, 608735/618640) and the highest is 99.67% (both 23andme, 636379/638469 and FamilyTreeDNA 718097/720477).

You can convert these roughly into centiMorgans using the table from ISOGG:

https://isogg.org/wiki/Autosomal_DNA_statistics

It's roughly 3370 cM. Essentially if is a total match within error.

---

Thanks, but table on that linked page makes no mention of concordant and discordant SNPs, and says nothing about how to convert from even "% shared" to cM.

Answer 8

A statistical approach to splitting up and amplifying snippets of the DNA that makes it to the lab, followed by various algorithms to map that to what the algorithms learned from varying other people's history...  

The origins are informative entertainment, though mine have been not too far off.  Then again mt FTdna and Ancestry have my self-match only a little better than either matches my sister...

In my case an unexpected match did find a grandfather on ancestry and it fit an anomalous origin that did not make sense under family lore.

Comments

That makes a lot of sense as to the variations; they're amplifying different parts of the ladies genome and then comparing against a set of people believed to be from this or that population.

I wonder how long it will take before they are able to hone the data pool to accurately reflect folk's origins, if ever.

---

They never can, in principle, because you don't get 25% from each grandparent.    So if one grandparent is Italian, but you only get 15% of your DNA from him, your DNA is only 15% Italian, however accurately measured, whereas on paper you're 25% Italian.

So which is your true ethnicity?

This is not something that goes away with better technology, better algorithms, bigger databases etc.

Answer 9

Here is my husband's theory - and a more scientific link at the bottom addressing identical twins with differences.

My grandmother is 1/4 Mvskoke Native American. I have even found our ancestors on the Daws rolls. Her mother was creek and Irish.

Also, my grandfather was German and English.

Interestingly enough,23andMe says French, English, and Scandinavian on 23andMe.

Ancestry.com says England, Ireland, Germany, Norway, Sweden (Scandinavian) and MyHertitage says English, Eastern European, and Ashkenazi Jewish.

Not a single one has the Native American Ancestry. Only one says German (Eastern Europe doesn't include Germany).

My husband explained it like this. DNA is like marbles. You have 46 marbles (chromosomes) You get 23 marbles from each parent. Now, imagine that each heritage (Native American, Germany, etc) is a different color marble. Each of your parents has 46 Chromosomes. Nature reaches in and randomly grabs 23 marbles from each parent and throws them into the soup. Depending on how far back the heritage is, there will be fewer marbles of that color, so the chances of getting a marble in that color drop. So, while I have Native American Ancestry (which I can prove), I didn't get the Native American DNA.

Also, DNA is interpreted differently between different sites. For example, when you look at the overlaps of Ashkenazi Jewish on MyHeritage, it colors in the regions for Eastern Europe - and Germany - The piece of ancestry that 23andMe calls French/German, but predicts French - and the part that Ancestry.com call German - but doesn't mention Jewish.

I bring that up because each site is testing my DNA, which doesn't change and should be the same across the board, right? It really comes down to computer interpretation. Here is an interesting article I came across when researching why my DNA results were not identical addressing differences in Identical Twins:
https://genetics.thetech.org/ask-a-geneticist/same-dna-different-ancestry-results

Comments

I like that article! I had speculated, above, that "perhaps it's just that the measurement of the DNA isn't 100% perfect." Sounds like that was exactly right. Thing is, their ancestral percentages match a lot better than the results I've seen of siblings, but this explains why it's not EXACTLY the same for them.

Basically, then, it's saying (and I believe this is really how it works) that if you tested YOURSELF a second time, with the exact same company, that you would see a similar variation. You would not match YOURSELF with complete accuracy.

---

So the twins thing is just a gimmick. They could have just used duplicate samples from the same person.

But it understates the case.  They'd get bigger differences using ordinary siblings, although of course they share the identical paper tree just as much as the twins do.

---

As far as ethnicity estimates go, one simple step is all it takes to disclose the...fluidity of the results. Take the exact same set of raw autosomal DNA data and upload to different sites...say, take an Ancestry test's results and upload them to FTDNA and MyHeritage. You will not see an identical evaluation of ethnicity even though the data are line-by-line identical.

The allele values at specific loci are all absolutely the same. But you won't get identical ethnicity results because those results are interpretive and assumptive, driven by a particular database. The results aren't about actual reference sequence nucleotide values but about some mystical, never-disclosed, never peer-reviewed, proprietary, ethnicity database.

Sigh. If I had a dollar for every social media post or YouTube video decrying DNA testing inaccuracy based on differing ethnicity estimates, I'd be talking to Jeff Bezos about a merger...

---

Excellent point, Edison!

I guess you could even go further and observe how even that EXACT SAME data - at the EXACT SAME company - gives different results, if you consider the "temporal dimension". In other words, they had a big revision in how they do their estimates on AncestryDNA late last year - my 2019 AncetryDNA ethnicity has wildly different ethnicity percentages vs my 2017 AncestryDNA ethnicity percentages.

---

This was exactly my point, Edison. I uploaded my data Ancestry and 23andMe data to MyHeritage AND paid MyHeritage to have them do a test. It was interesting to see the different interpretations. However, I didn't really do it for the ethnicity (though that was fun). I did it because I am adopted and trying to make family connections and find out more about where I came from. I found that, on my biological father's side, I am related to a couple of Notable people - so that was interesting.

Answer 10

After spending a week in an advanced genealogy class on using DNA in genealogy research, the ethnicity reports should not even be considered. They are mostly marketing. Not only are the estimates faulty but there is a bit of random sampling going on that can cause differing results. The vast majority od DNA testers have bought into the ethnicity. What that does is gets people to test and eventually the real matches help solve problems.

The rest of the DNA analysis is more than accurate enough to use in genealogy. If you really want to understand how it can be used, don't bother reading popular press and social media where they get it wrong most of the time and read articles from reputable, peer reviewed journals like the National Genealogical Society Quarterly.

Comments

Indeed. It's kind of an "ethical paradox".

On the one hand, the product is routinely oversold - it's portrayed as giving accurate percentages, when it doesn't, and even the ethnicities can be somewhat off. The "vast majority" have "bought into the ethnicity" because that's what they've heard over and over, in the advertising.

On the other hand, other aspects of the tests aid REAL genealogical research - which can ultimately lead to people learning about their real ethnicity (and more) anyway.

It's unusual to find someone who has over-reacted to a DNA test, with real harm having been done. Even then, it's incumbent on the buyer to do some due diligence before they start accusing mom of fooling around because dad was Irish and they didn't come up as exactly 50% Irish in their ethnicity. Reasonable people ask a few questions before accepting such conclusions, and the answers to those questions are really easy to come by.

---

Frank, thanks for mentioning ethics. I find ethical issues with the way Ancestry (and others) market their products.There are also ethical concerns in using DNA in real genealogy. Some people do react badly to finding out about misattributed parentage. It isn't always the test taker. Sometimes it is a parent/grandparent who didn't want it known. As genealogists, WE have to be aware of the ethical issues and not do harm.

---

I try not to be downright callous, in term of the "ethical concerns in using DNA in real genealogy", but it doesn't hit close to home for me (my lineage s DNA confirmed back at least to my gt-gt grandparents, with one explainable exception) and I have the mathematician's urge to "solve that problem", without regard to how anybody feels about it.

With that admission, logic is on my side. Suggesting that what is learned through DNA is questionable, ethically, is just "shooting the messenger". We expose the TRUTH, and I would assert that the descendants involved have the right - ethically - to know the truth about where they really come from, regardless of what kind of covenants were made by others (even when it involves the government - i.e., adoption cases).

The callous way I put it is, "If the guilty didn't want to be caught, they shouldn't have left so much DNA at the scene of the crime".  

So I don't really see ethical issues in what we do, as long as you're getting it right, and not carelessly jumping to conclusions. We seek - and often find - the truth. It's a question of how people choose to deal with it, and that is not for us to decide. We ARE obligated though - by rules of basic politeness - to deal with such issues with sensitivity and tact. It's not IMMORAL to tell a little kid that Santa doesn't really exist, but it WOULD make you a big jerk.

Really, with the general and steady breakdown in western society in recent decades, one wonders why NPEs even raise an eyebrow these days.

---

There have been suicides over the results of DNA testing, The ethicists believe that there are ethical issues involved. This mostly revolves around how information is revealed. It most definitely is OUR responsibility to act in an ethical manner and frequently our opinion on the matter isn't particularly reliable.

On NPEs, there doesn't appear to be much change in percentage over many generations. People just think it is worse now than it used to be.

---

To be clear, there isn't a damn thing that we could possibly uncover that would obligate ANYBODY to commit suicide over it. That's simply crazy behavior on their part, and in NO WAY our responsibility, regardless what so-called "ethicists" might say. We are NOT responsible for the emotional and mental health of others, and anybody who tells you differently probably has a stake in the aforementioned societal breakdown.

Not much change? You're kidding, right? In America, 3 of 4 births to black moms aren't illegit, and the white population is working hard to catch up. I suppose you could say that these don't necessarily represent "NPE"s, but it seems quite a stretch to claim that in this kind of environment - where what used to called a "scandal" is the norm - that the stats on that haven't budged. At best, you're saying that it's not that the dads aren't who were expected - it's just that nobody knew who they were in the first place, so not an NPE. Yippee. Doesn't take away from my point, at all.

---

NPE doesn't mean out of wedlock -- it means misattributed paternity (or, more rarely, misattributed maternity, such as in cases where someone discovers their biological mother was an egg donor). Basically, the daddy isn't who your mother claimed it to be. Multiple studies on the subject have shown that the NPE rate holds steady at about 2% of births. The outlier is Mexico, where a study found that something like 10% of births were NPEs. I don't suppose anyone really knows why -- maybe Mexican women are more likely to cheat, or more likely to carry a pregnancy to term even if they aren't sure who the father is.

Out of the thousands and millions of ancestors we all have, sheer probability means that we must have NPEs in our ancestry. Whether it's the classic 'Mom knew the neighbor a little too well', or 'Dad was sterile after the war so his brother did them a solid and knocked Mom up', or even 'Mom had a hot night of passion with a petri dish'.

---

"Multiple studies" say a lot of things about a lot of subjects. You're supposed to believe them unquestioningly, until new "multiple studies" come along that tell you it's 100% the opposite. Then you're supposed to believe THAT unquestioningly. It's scary how many people have no problem at all with following that pied piper, and would condemn you as "unscientific" for not following along like a lemming.

When considering such reports, the savvy consumer of such information recognizes the limitations and corruption within the scientific/academic establishment, and asks questions like (1) "How could they possibly do an experiment that really tells them that, with any reliability, without spending a vast fortune (which they undoubtedly did not spend)?" (2) "Could they possibly find out all the necessary information?" or (3) "Does the result make any sense at all?"

In this case, the answer to all these questions is a resounding "No."

Answer 11

The results are hardly mystifying when one puts just a little bit of study and thought into them. Most twins are not identical, and their DNA  should have differences.

Many people are reacting and claiming that genealogical DNA tests are not reliable, and that is demonstrably false. Y-DNA tests are very accurate in identifying the paternal line DNA of the testee, and thus is a very reliable tool for verifying the surname paper trail . . . if the DNA and paper trail do not agree, the paper trail is wrong. Autosomal DNA is very accurate in identify close family connections, reliable enough that law enforcement has begun realizing that it can be a tremendous tool for solving cold case crimes. MtDNA is less useful for most genealogists, but can be used to verify maternal lineage questions.

The one test use that every skeptic harps on is admixture analysis (ethnicity) of autosomal DNA. Doesn't seem to matter that every reputable DNA lab repeats ad nauseam that the admixture results are estimates, not hard facts. Estimates are allowed to be inaccurate. Admixture is a genealogical sub-study at best, and inaccuracy in this does nothing to diminish the reliability of autosomal DNA use in connecting families.

At present genealogical DNA can also be used to determine one other thing . . . race. Your results can very reliably confirm if you are homo sapien or an unknown alien species. Unless I've missed something in the scientific literature the test labs have a 100% success rate in correctly reporting these results.

Comments

Thank you John, for speaking sense. Personally, I was hoping for some Martian DNA but so far no dice...

---

Race has nothing to do with DNA as race is not a biological, scientific or genetic term. It has always been a socio political term.The American Association of Genomic Studies denounces that term as it has nothing to do with DNA. All human beings have the same alleles and due to environmental pressures, give rise to different frequencies of genes which in upturn give rise to the various population groups we have worldwide.  All these companies do is test a portion of your genome and determine if you match the various reference panel groups. Nothing more and nothing less. Also, none of these companies can tell a person about their ancestors’ ethnicities as they don’t know and can’t know how your ancestors identified themselves. Genetic genealogists have warned people about these two mistaken beliefs. All these companies do is provide an estimate of your ancestral makeup.

---

Anonymous, you missed the most important part of the race statement I made. The race that DNA 100% correctly identifies is the human race, which is the only true homo sapien race. Everything else is cosmetics.

Answer 12

You may find this article of interest because new research shows that even twins who are identical can have some differences.  

https://www.nytimes.com/2008/03/11/health/11real.html

For me this is similar to what we find but on a much smaller level with recombinant DNA.  These are not major genetic differences but they were just not expected to be there.   Nature has a way of fooling us and making us change our long held beliefs.  

Hope this article helps.