Zhou, Zhang, and Ebright found random nucleotide/base pair substitutions occurred in about one of every 27,000 pairs during PCR sample amplification. Some back of the envelope math: if the regions that 23 and Me et. al. are looking at are on average 700,000 pairs long, we can expect about 26 pair substitutions after the first round of amplification. But PCR is cyclical and any errors introduced in the first cycle get copied into the next cycle plus a new (on average) 52 substitutions on the resulting 1.4 million base pairs. I'm not sure how many cycles these outfits are using but I see 30 get bandied about pretty frequently. Add to that imperfect genetic profiling for certain parts of the world and you could see how such a situation might arise.
Mail-order genetic testing kits, which are all the rage right now, have been put through their paces by identical twins, and the results are a little baffling. These test kits collect your DNA, typically by you spitting into a tube, and then you have to send the package back to the manufacturers for analysis. The results are …
Saturday 19th January 2019 08:51 GMT David Harper 1
You're basically saying that every Ancestry/23andMe-type DNA analysis based on the product of PCR amplification is so full of errors introduced by the PCR process itself as to be useless. It's worth noting that the study which you cited by Zhou, Zhang and Ebright is from 1991, and focusses on a specific polymerase. The science of DNA-wrangling has moved on a LONG way since 1991.
Friday 18th January 2019 22:21 GMT Alistair
Monday 21st January 2019 21:45 GMT Michael Wojcik
It's perfectly reasonable to claim their first-generation (genetic) descendants can be expected have at least 25% of the mother's ancestry in each category.
"25%" because some of the supposed ancestry-marking genes that these snake-oil vendors are identifying might appear only on one chromosome of a given pair, and consequently have only a 0.5 probability to be passed on to a given offspring. We can assume that a reputable ancestry-determining organization1 would want a sufficiently large number of markers for any given ancestry characterization that we can expect a return to the average. In practice it should be closer to 50%; 25% is a worst case.
"At least" because the other parent might, of course, have a greater share of ancestry in a given category, boosting the child's.
Now, that might seem like a fairly weak claim; but it is a claim that such a service could report to a subject, about the subject's potential descendants.
1Which these are not. But that's why I wrote "can be expected": if that expectation is violated, it's just more evidence that these five firms are peddling rubbish.
Friday 18th January 2019 22:22 GMT HonourableTyr
Friday 18th January 2019 22:48 GMT Snowy
Monday 21st January 2019 11:16 GMT Loyal Commenter
Monday 21st January 2019 11:41 GMT AMBxx
Tuesday 22nd January 2019 08:57 GMT LucreLout
Re: Or maybe...
Oddly, it's unusual to see two identical twins who don't dress the same
?? I only know two sets and neither twin in each would be seen dead wearing the same clothes are their 'tasteless' sibling.
Same DNA. Different people. Sort of how you can build different things witht he same Lego bricks really.
Monday 21st January 2019 21:49 GMT Michael Wojcik
Re: Or maybe...
It's quite common for epigenetic changes and differences in health, diet, and environment to cause them to not be actually physically identical.
Indeed, it would be astounding if it were otherwise. Apparently a couple Reg readers 1) haven't gotten the news that genotype does not solely control phenotype (far from it); and 2) think they can determine whether two people "look identical" from a single photograph.
I don't know which is the more depressing observation.
Friday 18th January 2019 22:24 GMT HonourableTyr
Friday 18th January 2019 23:02 GMT the Jim bloke
Twins can have different growth and development through childhood nutrition and disease, and physical habits, and, well, there are these things called "shoes" that people attach to their feet, which have the effect of altering perceived height.
"picture of twins not showing what shoes they are wearing... your argument is invalid"
Saturday 19th January 2019 08:56 GMT DavCrav
"Perhaps they are fraternal twins who just look exceptionally similar? That is a possibility."
And, drum roll please, if these tests worked, they should still give the same answer for any person whose parents are the same. The fact that the article confirmed through a proper DNA test that they were identical notwithstanding, one daughter cannot be 13% from somewhere and the other 3% from the same place, if they have the same parents.
So we have two options:
1) Not only are the DNA tests from the proper place wrong, the mother wrong for thinking they were identical at birth, but also they must have different fathers, despite being twins, which might be a little bit tough, or
2) These tests are full of crap.
Saturday 19th January 2019 10:27 GMT Pascal Monett
Sunday 20th January 2019 19:16 GMT iromko
Re: 2) These tests are full of crap.
You've got it wrong: simple paternity test done in a lab in France goes for €150 because France has blocked other commercial services and allows peddling outdated technology from 90s for crazy price. That 99$ (during sale you could get it for 49$) test is several generations more advanced than that paternity test crap.
Sunday 20th January 2019 06:19 GMT Jordan Davenport
"And, drum roll please, if these tests worked, they should still give the same answer for any person whose parents are the same. The fact that the article confirmed through a proper DNA test that they were identical notwithstanding, one daughter cannot be 13% from somewhere and the other 3% from the same place, if they have the same parents."
While yes, the article confirmed the sisters actually are identical twins, what you stated is incorrect.
In the case of non-identical siblings, assuming meiosis completes correctly, each sibling inherits approximately half of their nuclear DNA from each parent, ignoring the discrepancy between the X and Y chromosomes. Since these functional halves are not identical between siblings, each sibling receives a different percentage of ancestry from any particular source, only guaranteed to inherit specific sets of genes in common. For instance, any two brothers will inherit practically the same Y chromosome, and any two sisters will inherit practically the same X chromosome from their father. As such, in the case of two non-identical sisters from the same parents, one daughter actually can be 13% from somewhere and the other only be 3% from the same place.
Taken to the extreme, though highly improbable, it is theoretically possible for two siblings (one genetically male and one genetically female) to be born sharing no common nuclear DNA whatsoever outside of what is common to our species, inheriting the exact opposite halves of each parent. In this scenario, it would be possible for one to have no common ancestry to the other in their nuclear DNA. The chances of this happening are so infinitesimally slim that we can reasonably assume that this never has happened and never will happen in the entirety of our species, but it is possible.
Sunday 20th January 2019 18:16 GMT Bill Gray
This point also has come up in discussion of US Sen. Elizabeth Warren's claims to have a Native American ancestor five generations back (i.e., 1/32 of her ancestry). She (foolishly, in my opinion) allowed Twatter taunts from the Orange Lord to persuade her to get and release a genetic test, which showed something more along the lines of it being 1/1024 of her ancestry. At the time, it was mentioned that even a properly "advanced" genetic test might miss a five-generations ancestor, for basically the reasons you suggest : you might get a good bit of DNA from that person, or nothing noticeable.
Monday 21st January 2019 01:24 GMT Anonymous Coward
Tuesday 22nd January 2019 00:18 GMT Fred Goldstein
She didn't say she did or didn't absolutely have native ancestry. She has however noted that her father's family did not approve of his marrying her mother (if I have the direction right) because she was part Indian, and the father's white family did not like that. So either the racist family of the father was wrong, or there was some Indian ancestry, which is pretty common among toubab in Oklahoma (f/k/a Indian Territory).
Monday 21st January 2019 09:24 GMT DavCrav
"While yes, the article confirmed the sisters actually are identical twins, what you stated is incorrect. [snip]"
Thank you for, in great detail, explaining why I was right after all. I guess my point, although apparently overlooked, is that the tests are not only wrong, in the sense that identical twins should give exactly the same answers, but also meaningless, in that siblings should give the same answer if it were possible to do what the product says it does.
Monday 21st January 2019 22:20 GMT Michael Wojcik
siblings should [receive] the same answer if it were possible to do what the product says it does
Particularly given the vague ("European ancestry") and imprecise ("13%") nature of the results, the tests would only be useful if they represented a fairly strong signal. (That is, the vagueness implies that test is neither accurate nor precise; thus no result should be reported unless the test finds it with strong statistical confidence.)
That means a reputable organization would only report ancestry components supported by a relatively large number of markers.
Given a large number of markers, it becomes extremely improbable that they wouldn't return to the mean; and so very improbable that full siblings wouldn't receive fairly similar results.
Consider: Assume the presence of some gene M contributes to a finding of ancestry X. Sibling A is found to possess M. Does sibling B possess it?
- At least one parent has at least one copy of M. There's a chance that parent has two copies, and (barring mutation) all sibling children will have it; call that p1. Otherwise, a given sibling has 0.5 chance of getting M from this parent.
- There's some probability that the other parent has at least one copy of M. Call this p2. If p2 is true, there's a p1 chance1 that parent has two copies of p, in which case all sibs have it. Otherwise, there's a 0.5 chance that a given sibling will have received M from this parent.
So what's the chance of B possessing at least one copy of M? Let p3 = 0.5 + 0.5p1, the probability of getting it from the parent known to possess at least one copy. Then B's chance of getting at least one copy of M is p3 + (1 - p3)(p2)(p3), or p3(1 + 1 - p3) or p3(2 - p3). Note p3 is almost certainly > 0.5. So B's chances should be better than 0.75.
Unless I've messed something up in my calculations, which is entirely possible.
We assume p is fairly common among people with significant X ancestry - that's part of what makes it a good marker for ancestry. (We also assume it's significantly less common for people without X ancestry.) So if A and B's parents share significant ancestry - which is more common than not - then sibling ancestry results should converge even more strongly.
1p1 as well here because we take p1 to be the probability across the population of all people who have at least one copy of M that they have two copies of M. Obviously within that population it's likely there will be subgroups who have greater or lesser probability of having two copies, but we don't have any information to partition at that level.
Monday 21st January 2019 11:21 GMT Loyal Commenter
And, drum roll please, if these tests worked, they should still give the same answer for any person whose parents are the same.
Actually this is not true. You inherit half of your genes from one parent, and half from the other. Which genes you get from which parent is entirely random, so if they are looking for a specific genetic marker that one parent has, you have a 50% chance of inheriting it. Siblings will inherit different genes to each other, although statistically, if you are looking at enough markers, you would expect to get results that are very close.
My suspicion is that these tests are not looking at the whole genome, but sections of it. They will have catalogued variations in several specific sections of DNA, but aren't looking at exactly the same sections in each test they do. This would account for the variation in results. For example, one test may look at several fragments, lets call them A B C D E F. One run picks up A B D and F, the next gets B C D and E. The results are therefore going to vary.
However, as they get more and more samples, the statistical confidences will get better, and the overall ancestry results derived from analysing different framgents are likley to converge.
Saturday 19th January 2019 20:55 GMT Anonymous Coward
Monday 21st January 2019 15:22 GMT anothercynic
If a professional DNA lab does an analysis and finds them to have identical DNA (as one should as an identical twin) but 23andme et al show wildly varying results, I would trust the professionals, thank you very much.
For any avoidance of doubt, an organisation like Sanger I would trust (given that they worked hard over many years to sequence the entire human genome with exceptional accuracy)... anyone who claims to give you accurate results in days from a (dodgy at best) spit sample is lying through their teeth.
Friday 18th January 2019 22:25 GMT Anonymous Coward
Friday 18th January 2019 22:35 GMT Eddy Ito
Monday 21st January 2019 07:01 GMT Doctor Syntax
"Most of various DNA companies allow you to download the genetic data they have."
TFA has a link to https://www.cbc.ca/news/technology/dna-ancestry-kits-twins-marketplace-1.4980976 Let's take a look:
"The team at Yale was able to download and analyze the raw data set that each company used to perform its calculations."
"According to the raw data from 23andMe, 99.6 per cent of those parts were the same, which is why Gerstein and his team were so confused by the results. They concluded the raw data used by the other four companies was also statistically identical."
I do, however, wonder about "identical" when qualified with the word "statistically". I'd have been inclined to say "indistinguishable" - and"were" rather than "was".
Friday 18th January 2019 22:52 GMT Anonymous Coward
So, let's do a quick analysis of this non-story. Two people who are to quite clearly non-identical fraternal twins have gotten confirmation from several different companies that they are non-identical fraternal twins. They are so unwilling/unable to accept the results that they just rubbish all the companies and tests involved.
From my own personal experience submitting samples by different family members with different surnames in different countries the relationship predictions, out to 3rd cousins, has been 100% accurate. I can confirm this through having contact with a large extended family across said countries who all still keep in touch.
As non-identical fraternal twins they can end up with very different proportions of genes from different geographical ancestry from each parent. You only share 50% of your DNA with each parent. With a non-identucal sibling you'll share about 25%. This is of the variable regions.
This article just shows a total lack of understanding by the author and by the originating source.
Saturday 19th January 2019 00:26 GMT Anonymous Coward
Now let's analyse your non-post for a moment. Maybe you only skim-read the article? I say that, because this line seems to make your post redundant..
" Boffins at Yale university, having studied the women's raw DNA data, said all the numbers should have been dead on.
"Despite having virtually identical DNA, the twins did not receive matching results from any of the companies," the investigative team noted in their report this week."
Saturday 19th January 2019 05:31 GMT CheesyTheClown
Boffins or Bafoons?
They obviously are different :)
Journalists generally have absolutely no respect for science.
You're quoting "Boffins at Yale university, having studied the women's raw DNA data, said all the numbers should have been dead on."
Let's start by saying that the definition of a pair of identical twins is that they were both hatched from the same egg, or more accurately the egg split in half after being inseminated and produced two separate masses which eventually developed into two individual humans. I have not looked it up and I'm already guilty of one of the same critical mistakes made by the journalist which is that I have no verified my facts. But this is how I understand it.
If I am correct, then cellular reproduction through mitosis should have split the nucleotides of the original cell precisely. This means that over a certain percentage of the genetic pairs to be reproduced survived in tact. Let me clarify, from what I can fathom, simply mathematical entropy dictates that there must be an error in every cellular reproduction. It is not mathematically possibly for two cells to be 100% alike. 99.9999% is realistic, but not 100%. This is a mandatory aspect of science. To make the particularly clear, refer to Walter Lewin's initial lecture from Physics 801.x in MIT Open Courseware on Youtube where he explains how to measure in science.
So, we're presented by your quote "Boffin's at Yale..." which leads me to ask :
- What is a boffin?
- What is the measure of a boffin?
- Who is qualified to measure whether this individual is a boffin?
- What is the track record of accuracy by the boffin?
- Was the boffin a scientist? If so what field?
- Was the boffin a student? If so, what level and field?
- Was the boffin an administrator? Were the a scientist before? When did they last practice? How well did their research hold up when peer reviewed? Did they leave science because they weren't very good at it and now they wear a suit?
And "having studied the women's raw DNA data"
- How contaminated was the data (0% is not possible)
- How was it studied?
- Were all 3 billion strands sequenced and compared? Was it simply a selection?
- Did they study just the saliva as the companies did or was it a blood sample?
- Does saliva increase contamination?
- Was the DNA sample taken at Yale?
- Was the DNA sample shipped?
- If it was shipped, was it shipped the same way?
- Could air freight cause premature decay or even mutation, etc...?
And "said all the numbers should have been dead on"
- Was the boffin really a boffin? (see above)
- What does dead on mean? What is the percentage of error?
- What numbers are we talking about?
- Should the results been identical between the twins?
- Could two samples from the same twin produce the same discrepancies?
- Could two separate analysis of the same sample produce the same discrepancies?
- What happens when one person spits into a tube and then the spit is separated into two tubes?
- Did the boffin say "all the numbers should have been dead on" or did they provide a meaningful figure?
- Is this the journalist's interpretation of what the boffin said?
- Are these words the words of the original journalist or was it rewritten to make it sound more British? I've never heard any self respecting person use the term boffin if they actually had a clue to begin with. Same for the word expert.
- Did the "boffin" dumb down the results for someone who is obviously oblivious?
Overall, does this comment translate to :
"Qualified scientists with proven track records specializing in the field of human genetic research as it relates to ancestry evaluated 98% of the sample from each of the two twins, verified that they are in fact identical and that there should be no more than 5% margin of error when comparing the results of genetic studies between the two girls."
I'm not a scientist. I'm barely a high school graduate. But before I would dispute the accuracy of what the AC said by defending it by stating "Boffins at Yale university, having studied the women's raw DNA data, said all the numbers should have been dead on.", I would absolutely start with attempting to answer those questions above.
At this time, while I believe strongly that these twins are in fact identical and I believe they were verified at some point in their life (whether by a "boffin at Yale" or elsewhere) to have been conceived from the same egg, I would not offer it as evidence without further research or conclusive proof. That would be an insult to science. My believe is irrelevant unless this were some high school sociological report for a political science class. And even then, that's a misnomer.
Let's address ancestry as well.
I read the results and let's take an except as you did.
- 23andMe reckoned the twins are about 40 per cent Italian, and 25 per cent Eastern European;
- AncestryDNA said they are about 40 per cent Russia or Eastern European, and 30 per cent Italian;
- MyHeritageDNA concluded are about 60 per cent Balkan, and 20 per cent Greek.
I'm no expert on ancestry, but I have questions.
- What does it mean to be italian, eastern european, russian, etc...?
- Within how many generations would the be counting?
- Would the 40% Italian refer to the 1800's very likely means balkan and greek in 200AD?
- Is Russian of east european or east asian or central asian descent?
- Is balkan east european or russian?
What I'm reading here is that all three companies were in 100% agreement. If anything, for an imperfect science, it's impressive how perfectly they agree.
As another item
- Two of the tests reported that the twins had no Middle Eastern ancestry, while the three others did, with FamilyTreeDNA saying 13 per cent of their sample matched with the region.
I'm pretty sure that if we believe modern science, everyone on earth should come from the middle east and before that Africa. So unless the twins come from another planet, they have middle eastern descent. The question is, how many generations back? Also, what counts as middle eastern descent? If we refer to religion, only a few thousand years ago, Jerusalem was in war with the Seleucid empire which almost definitely spread seeds from the middle east to the Mediterranean or maybe the Mediterranean seed was spread widely enough to influence what is considered the middle eastern bloodline today.
Again, I don't see the data conflicting, I would need far more information to sound even moderately intelligent.
And I'll attack one more
- On top of this, each test couldn't quite agree on the percentages between the sisters, which is odd because the twins share a single genetic profile.
This is absolutely 10000% not true. They are born from the same egg. I can't find her birth date, but I'd put her at about 33 years old, but that may simply be because she dresses like an old lady. Either way, let's round to 30 years old.
Unless she and her sister have been in the womb together until last week which I doubt, they should have been through an infinitely different life causing infinite changes in their their genetic code from one another. Their genes can't possibly be that similar anymore. Almost certainly less than 99% similar now. That's at least 30 million differences from one another given that a human gene consists of 3 billion nucleotides or genetic pairs.
Consider that simply walking in the sun causes genetic mutation. Pressures from drinking water can cause genetic mutation (I read a peer reviewed paper on this, but can't cite it).
So, please don't bash the AC, he's as full of shit as I am... the only thing I got out of this article is that some girl who has a twin sister has now made an international impact with her ignorance of science in an effort to make a headline to increase the ratings of her TV show which should have served the purpose of informing people of facts.
I would like to see some real research on this topic from people who are far smarter than me. At this time, I have many questions and wouldn't even know where to start answering them.
Saturday 19th January 2019 12:09 GMT Anonymous Coward
Re: Boffins or Bafoons?
If you are going by what you perceive as right (you THINK they don't look like twins), rather than the article telling you they are twins.
They are identical twins, because I, unlike lazy you (I don't know if I can call you lazy, you wrote what, a 10,000 word post?), also read articles outside of the register, other sites tend to carry more info on the boffinry bit.
They had their DNA sequenced by Scientists at Yale, at the start of the progam, to create a baseline before they sent tests off to test the validity of these devices.
Rather than backing up the AC with a monster post full of even more crap than the original poster, why don't you just read the article on another site?
Really saves you looking a complete idiot.
Sunday 20th January 2019 04:09 GMT CheesyTheClown
Re: Boffins or Bafoons?
I followed up on more articles on this now. And while you have some valid points, there are still many questions left open and yes, they "dumbed the shit down" for news and TV substantially. Multiple times during the interview the "boffin" damn near back peddled in order to keep the words small enough for the journalist to understand.
The people who were interviewed were clearly "boffins" in the field. This means they were "experts" on something and not on another, but the people who interviewed them didn't thing to talk to someone who could interpret the data more clearly.
A boffin in my experience is someone that dumb people call smart without understanding what that means so they can tell other dumb people that they should trust this data because it comes from a genuine boffin. For example, a guy in a blue shirt at an electronics store with a logo that says "Geek" is clearly a boffin, he has a shirt to prove it. Imagine how impressed those same people are by the geniuses at the genius bar.
I believe the "boffins" in question here (after reading their credentials) are specialists on topics relating to genetics but lack interest or focus on more commercialized genetics (meaning mail order) and genealogy. They were questioned though as if this would be a particular area of expertise for them. This means that although they might be able to speculate in areas they lack the data set to speak authoritatively on, they answered anyway since they didn't realize how wide spread their findings would be published in the mainstream media.
The numbers published leave most of my questions in tact.
- How does shipping impact it
- Where the results actually contradictory?
- What were the results of two samples from the same twin from the same lab?
I can go on.... the point is, the people interviewed were geneticist and I assume pretty damn good ones. They are experts in sequencing and applying their findings to medical application. I'm sure there are even people with an interest in forensics on the team. But this team doesn't seem to have a whole lot of background on historical migratory patterns. They also don't appear to be accounting for shipping methods which means they are probably incredibly brilliant people who when handling samples minimize contamination and they're being asked to evaluate how a sample stuck into a non-sterile device and then shipped through random methods would fair.
You're right, I should have read more and now I have. Before looking like an idiot and simply jumping on "boffins at yale" as an excuse to grandstand, I would still need many questions answered.
Cool... they are identical and proven.
- What does that mean in context of a pair of 30 year old identical twins who clearly show substantial variations in development?
- What is the margin of error when taking two samples from the same twin vs one sample from each twin?
- What is the result when ensuring the labs receive non contaminated samples?
- When the results are interpreted by a historical anthropologist and accounting for migratory patterns and possible differences introduced by how generational data is interpreted, were the results actually different or was this a rounding error?
But... I guess since now that you and I have read the same articles and seen the same interviews, the fact that you seem to be happy and I still seem to want reduce a great number of variables... it could be that I'm an idiot or it could be that I have high standards of what I consider to be meaningful data. If I had interviewed the "boffins" I would have asked for tolerances and certainties for all measurements.
You can tell the scientists involved here intentionally kept the words small and simple for people like you to avoid confusion like this. They even came straight out and said things like "identical". Absolutely no scientific evidence ever has been admissible without defining the precision and 0% deviation is not a possible measurement. 10^-100 percent is, but 0% isn't. Any scientist presenting information in any way without that is trying to sooth the fools. And the fools will quote him or her without asking additional "boffins" additional questions to multiply the tolerances to reduce the margin of error of their findings.
The AC was crap... your response was arrogant and uniformed crap. My post was arrogant uninformed crap and the story was misleading crap. The difference is, I'm claiming we're all full of crap and you're choosing which line of crap you'll present as fact to degrade people.
Tell me. If the article said buffoons instead of boffins, would that margin off error impact how you'd use the "evidence" to show you're superior enough to call people idiots?
So, are you prepared to step up and admit you are full of shit too or will you continue to present yourself as the "boffin" to we lowly idiots?
Sunday 20th January 2019 15:13 GMT Richard Boyce
Re: Boffins or Bafoons?
I gave you a thumbs up because you're making people think and raising quality standards by playing devil's advocate.
If a lawyer defends a well-known criminal, the lawyer may strongly suspect that his/her client is guilty, but it would be in *everyone's* interest that the evidence be tested very carefully.
Monday 21st January 2019 07:01 GMT Doctor Syntax
"This article just shows a total lack of understanding by the author and by the originating source."
Did you read the originating source? Did you notice the bit where the downloaded 123andMe data had 99.6% agreement? Can you, from your alleged understanding, explain how closely the parents would have to be related to each other to produce a result like that from non-fraternal twins? I don't think even the Hapsburgs could have managed it.
Monday 21st January 2019 07:04 GMT diodesign
Re: anonymous coward
"This article just shows a total lack of understanding by the author and by the originating source."
You haven't read nor understood the article(s) - the sisters are identical twins. Their DNA profile is 99.6% the same. Each test should have returned the same percentages for them.
Monday 21st January 2019 12:38 GMT Doctor Syntax
Re: anonymous coward
"Each test should have returned the same percentages for them."
Not so much each test, more the interpretation of the tests. The 0.4% difference (presumably experimental error in sequencing and/or mutations during development) seems to have thrown the interpretive algorithms of some vendors more than others.
To my mind the differences in interpretation between the other two vendors is interesting. They give very similar results between the twins but pretty substantial differences between the vendors in what these mean. It suggests to me that the real problem in all of this stuff is the data on which the interpretations lie. Any place in Europe has a population whose ancestors got there by a variety of different routes and which swapped elements of the population with other places. The result isn't entirely homogeneous but it's certainly fairly mixed and each of these businesses (and academic investigators also) are taking sub-samples of this mix and characterising them as some sort of type specimens and then using them to interpret subsequent analyses. I doubt the underlying data are really good enough to make proper sense out of this, nor will they be for a few more years.
It all reminds me of the situation back in the '60s/early '70s when we were starting to apply carbon dating to archaeology (and palaeoecology) but without a good enough corpus of dating material to make real sense of it. E.g. having some nicely dated pollen diagrams of the Irish late bronze age and a nice archaeological write-up of the Irish late bronze age which seemed to fit but without a single carbon date for the archaeology to test that fit.
Friday 18th January 2019 23:33 GMT knelmes
Friday 18th January 2019 23:36 GMT hitmouse
Re: "Fall Creators Update"
The main people interested in having genetic testing to trace ancestry are those living in nations mostly formed by immigrants: USA, Canada, Australia etc. Thus the pool of information available to testing companies about gene distributions in many countries doesn't reflect "native" populations of those countries.
Saturday 19th January 2019 00:05 GMT FF22
Monday 21st January 2019 15:15 GMT theModge
Re: Recipe for success
I don't know; you could do better than that with a bit of marketing:
1. Get Samples
2. Look at origin post mark.
If American, tell them they're Irish with a dash of Native American. I swear pretending they're Irish is the American national hobby and a touch of Native American makes you seem interesting and special.
Northern UK? Tell them they're Vikings
Southern UK? Tell them they're related to the Plantagenet royal line
I'm sure there's some others I could offend with a bit of research.
3. Look at surnames. If they give you an obvious clue, make it up on those grounds
4. Still be more accurate than the existing players
5. Give all the money to company directors and declare bankruptcy before people sue you for fraud.
Saturday 19th January 2019 04:05 GMT Random Q Hacker
Saturday 19th January 2019 06:02 GMT Anonymous Coward
I think the term "Microchimerism" or "Mosaicism" is the technical description and yes it can lead to very strange results on a genetics test.
Interestingly it has emerged that "lost twin syndrome" is more common than previously believed, with evidence that a lot of people were once twins in utero because the appropriate check wasn't done on the placenta.
A little while back I read of a case where maternity (!) was disputed because the mother unbeknown to everyone had mosaicism and by chance the sample they took had one cell type primarily and the other was discarded as noise.
Monday 21st January 2019 07:00 GMT Brewster's Angle Grinder
Re: Re. Chimeras
Mosaicism is different to microchimerism. "Mosaicism" or plain chimerism is where two zygotes have fused in utero and organs develop with differing genotypes: you can literally end up with one arm with one genotype and the other arm with a different one, or with two different blood types; the case you're referring to (Lydia Fairchild) was an example of this. Microchimerism is where there are a few stray cells with a different source; mothers, for example, often have cell lines from their children lodged in their body or floating in their blood.
And. for the record, I think we can rule out both. Spitting in a test tube or swabbing a cheek wouldn't spot microchimerism. And even if the experts hadn't looked at the twins' DNA, what are the chance of two eggs being in the womb, one of which splits into two foetuses and then one of these foetuses fuses with the fraternal egg in such a way that the surviving children look identical but have different saliva glands?
Saturday 19th January 2019 08:02 GMT S4qFBxkFFg
What had they been eating?
These things work with saliva swabs, right?
Does this account for the fact that there is DNA in the things that we eat?
Now I'm not suggesting that someone has received a result claiming they're 5% lettuce or something, but having a small amount of DNA from your last meal, or even just traces of saliva from kissing a partner, could affect the tests in unpredictable ways.
Saturday 19th January 2019 09:44 GMT FelixReg
There are several misled comments here.
The DNA tests don't confuse fraternal and identical twins. Therefore, forget whether you think these two women are identical. They are. Or are being purposely misleading.
The DNA outfits report around 600,000+ "SNPs" - pairs of molecules (one from each parent) that are of 4 types, labelled "A", "G", "C", and "T". The DNA companies choose their 600k+ SNPs from our 3.5 billion SNPs. The ones they choose vary in A, G, C, and T value between different humans. Almost all of our 3.5 billion SNPs are the same for all of us.
Each company chooses a different set of SNPs to report. 23andMe has reported at least two different sets of SNPs, depending on when you did their test. I've found Ancestry,com and 23andMe tests have from 100K to 300K of the same SNPs between any two test/versions (out of the 600k+ each test reports).
Two tests for the same person vary by only a handful of SNPs. That is, there is noise, but it's not very loud. Different companies' tests report different SNP values, too. Again, very quiet noise.
The serial killer thing in the article used these DNA kit results. Crime labs use something quite different.
"Ancestry" or "ethnicity" is calculated using fancy arithmetic. Different companies do, indeed, report different "ancestries". As would be expected. Issues, among others:
1) Different SNPs between companies and company-versions.
2) Ancestor? When? These guys generally are shooting for a few hundred years ago in Europe, not thousands. But outside Europe? Read on.
3) How much (if any) data backs up "ancestry"? You can't find what you don't know. These outfits have SNP combo examples from south-south-eastern Duchy of Euroland, but have bupkus for much of the world. Only recently have they started differentiating Siberia from Chile, for instance. And Siberia and Chile diverged over 10 thousand years ago!
4) Ancestry calculations depend on knowing which AGCT of the SNP pairs go to which parent. That's rather a trick to know when you don't have DNA data for either parent. The article doesn't say whether these women got their mother or father to do a test so such information could be nailed down.
The FDA doesn't care about "ancestry" reports. They "FDA'd" 23andMe a few years ago because of health reports. 23andMe has since jumped through the FDA hoops and now benefits from the usual lack-of-competition regulation causes. Don't hold your breath waiting for useful health information from these DNA tests.
I'm Scandinavian to two of the DNA outfits, Brit to one of them, and changed from Scand to Brit at a fourth outfit a year ago. Smaller percentages change often. These guys are busy tweaking their code and data.
23andMe, for one, allows you to choose how reliable you want their ancestry information to be. The article doesn't say anything about the settings these two women used. If you want 23andMe's most reliable report, don't expect a lot of specific ancestry information. "Northern Europe", folks. Or "African". Sort of at the level our eyes see without close examination.
Hope this helps.
Saturday 19th January 2019 12:59 GMT Wellyboot
Re: No surprise
I'll go with that.
Picking a couple of European countries for 'origin%' is just plain silly, Movement within the Roman empire 2000 years ago (and all the intervening history) has pretty much averaged out the whole European population. The far corners (Balkan, Baltic, Iberian) will show some small difference to nudge the results around a bit but no further than the visually obvious.
Saturday 19th January 2019 13:55 GMT FelixReg
Re: No surprise
@Wellyboot : Well, "silly" is in the eye of the beholder. How about entertaining? As evidenced by the number of people who have plunked down $50 to $100 to be entertained.
There certainly are objective DNA differences around Europe. And, if your location-DNA info is from recent enough, you're likely to have cultural idiosyncrasies from your DNA location. That can be interesting to ponder.
As others have mentioned, a lot of the interest in DNA ancestry comes from "immigrant nations". Where did those old folks come from? For instance, ancestry can be particularly interesting to Negros in the Western Hemisphere whose families go back to slave ships. What other way than DNA is there to get a line on where your ancestors lived in Africa? Sure, we're not talking about particularly useful information (at this time), but that doesn't stop a person from being quite curious. Those various African source locations were a loooong way apart geographically and DNA-ly. Curiosity! Which one(s) did your folks come from? Makes history and geography come alive.
Monday 21st January 2019 07:00 GMT Doctor Syntax
Re: No surprise
"Two tests for the same person vary by only a handful of SNPs. That is, there is noise, but it's not very loud. Different companies' tests report different SNP values, too. Again, very quiet noise."
Agreed. It must be the noise sensitivity of the "ethnicity" interpretation that accounts for the apparent difference between the sisters. In fact the major components are fairly close for any given pair of analyses. It's the smaller components which seem to be beneath the noise floor so to speak. Ancestry and MyHeritage seem to come out better on this basis.
The real problem is where the components get a geographical label attached. Humans have been wandering back and forth for some considerable time. Take any particular person from any particular place in Europe or round the Mediterranean (which seems to be what these figures are suggesting) and you'll have picked someone whose ancestors will of necessity have taken a long route - and a variety of long routes - to get to that place. Take a second person from some other place and their ancestral routes will also be long and varied and some of them will coincide with the first person's.
Add that up over the the European and Mediterranean area and then reflect how much sense it makes to try to work out what Greek or Italian or Balkan or any other of the labels means. It's no surprise that although Ancestry and MyHeritage can match the twins to each other within the noise limits they differ wildly from each other in what those mean when they try to sort out Italian from Balkan from Greek from Eastern Europe from Russia.
My ethnicity? 90% sceptic, 10% others.
Saturday 19th January 2019 10:38 GMT David Roberts
How many people base a crtique of genetic sequencing on a single photograph of two people with different hairstyles.
Perhaps they checked with a man in the pub for a second opinion?
Speaking as someone who is regularly mistaken for a twin when my brother is a year and five months older than me. Could have just been an exceptionally long pregnancy carrying the second twin, of course..........
Seeing is believing.
Saturday 19th January 2019 11:34 GMT HKmk23
Saturday 19th January 2019 12:20 GMT Marketing Hack
These mail-in genetic tests are crap
I have a 2nd generation Chinese-American friend who took one. Both his parents came from Guandong/Hong Kong and one of them can trace her lineage pretty accurately because the family have been merchants and small businesspeople in the same town for the last 300 years or so (that they know of).'
His test came back showing that he was only 50% Chinese.
Saturday 19th January 2019 12:22 GMT chivo243
Wow, seems that the down voters are out in force with blinders on to boot. El Reg reports a "Soft Journalist" has put forth a sensational claim, tearing down a novelty product at best, backed by a grad student at Yale? Before anyone jumps on that, do you think anybody making any kind of money with a reputation at Yale would touch this project? Then it seems quite a lot of readers-commenters have drunken the koolaid? There are too many steps where the wool could have been pulled over the facts...
There are so many questions that need answers before anyone should take this as science. It's sciencey-ish infotainment. If it quacks like a dog it must be a platypus.
Saturday 19th January 2019 18:40 GMT ibmalone
Monday 21st January 2019 10:40 GMT Anonymous Coward
Not exactly harmless infotainment though as they result in building up a genetic profile of your relatives too.
and I personally would be VERY surprised if the slightly more accurate results of these individual home DNA testing kits did not somehow get accidentally backed-up in the Prüm extensions database, the UK NaDNA 'base, the FBI 'base, KGB etc etc
Saturday 19th January 2019 17:12 GMT Phil Endecott
Interesting story on the BBC anout how quite large numbers of people have been bought these tests as presents because they were interested in their family history, and then discovered their parents weren’t who they thought they were. Cue very elderly parents who thought they’d take their secret to the grave getting a nasty surprise.
Sunday 20th January 2019 00:54 GMT Anonymous Coward
Sometimes you dont need a test
Had a personnel problem. One of my guys, a very pale Scandinavian type, had just about gone postal at the hospital when his son was born. His wife is Irish and can trace the old family back many generations. Baby? Very dark, with some resemblance to her work buddy.
Sunday 20th January 2019 18:46 GMT Anonymous Coward
Re: Sometimes you dont need a test
'..His wife is Irish..'
'..Baby? Very dark...'
Hmm, there are a number of the 'Black Irish¹' whose skin tone is rather dark (e.g. my McKinnon relatives, and I've seen it skip generations there and even throw up an albino)
'..with some resemblance to her work buddy.'
Ah, never mind then..
I will say I always marvel at how people can tell these things looking at a newborn, I've never been able to summon up even a minor degree of the physiognomancy I've seen others display after a cursory glance at the face of a 4 hour old infant (even after being told 'he has his father's eyes etc. etc I still can't see it.....mind you, there's always the chance that I'm not doing the proper comparisons and that the physiognomancer is more 'in the know' than I am..)
¹Aye, I know it's a term whose origins are somewhat contentious..
Monday 21st January 2019 09:34 GMT james_smith
Re: Sometimes you dont need a test
PIgmentation in new born babies can be very variable. My brother was blonde haired and blue eyed at birth, but slowly changed to brown hair and eyes. I was dark skinned at birth, which was initially put down to the Romani ancestry on my father's side, but I became very pale skinned within a few months.
Monday 21st January 2019 13:58 GMT deadlockvictim
The Black Irish
If you want to find the black Irish, go up to Norn Iron around Marching Season. And while they may not be too happy at you calling them Irish, they are very proud of being black.
Otherwise, Ireland has some black sons of whom she is very proud — Phillo being one of them: https://www.youtube.com/watch?v=d2OcIqwmSaY
Sunday 20th January 2019 06:42 GMT Barry Rueger
Sunday 20th January 2019 06:50 GMT WatAWorld
FDA approved blood glucose meters diabetes depend on for their lives only claim accuracy +/- 10%.
Testing the services could more accurately have been done by submitting samples from the same person twice than using twins you think are identical.
FDA approved capillary blood glucose meters that diabetes depend on for their lives only claim accuracy +/- 10%.
+/- 10% despite displaying results with a grossly misleading 2 digits.
You can use google to check up on the accuracy of scientific full sequence DNA tests while a bit higher, it omits large stretches of the full sequence it was supposed to analyse.
And "broadly european" includes French, Germany, etc. It is one of those,as precision goes up accuracy goes down. When they use the limited data in their database to try to narrow regions down more the number of times out of 20 they're right drops.
I'm in Canada and I read the CBC result. And I've used 23andMe myself. Yeah, if you ignore all the cautions and want to use the test for something it doesn't claim to be, you're going to be wrong.
But unlike with FDA approved +/- 10% lab test results, nobody dies.
Normally you'd want to repeat a test with suspect results, to see if it was a hard or soft error, to see what went wrong. The CBC did not do that.
I suspect that what went wrong with the 23andMe samples, the ones that had significant errors was that either:
1. One or both samples was contaminate when taken or when opened.
2. One of both samples was damaged in transit, perhaps by high or low temperatures.
3. An old database was used to analyze one sample and a new database used to analyze the other. That could explain the older less specific output.
Sunday 20th January 2019 06:54 GMT WatAWorld
Re: FDA approved blood glucose meters diabetes depend on for their lives only claim
The original CBC article quoted is here:
US National Institutes of Health report on the cost and inaccuracy of human genome sequencing in academic and medical settings
"Another important driver of the costs associated with generating genome sequences relates to data quality. That quality is heavily dependent upon the average number of times each base in the genome is actually 'read' during the sequencing process.
During the Human Genome Project (HGP), the typical levels of quality considered were: (1) 'draft sequence' (covering ~90% of the genome at ~99.9% accuracy); and (2) 'finished sequence' (covering >95% of the genome at ~99.99% accuracy). Producing truly high-quality 'finished' sequence by this definition is very expensive; of note, the process of 'sequence finishing' is very labor-intensive and is thus associated with high costs.
In fact, most human genome sequences produced today are 'draft sequences' (sometimes above and sometimes below the accuracy defined above)."
Monday 21st January 2019 12:38 GMT Doctor Syntax
Re: FDA approved blood glucose meters etc ( el Reg makes title too long then complains)
"I suspect that what went wrong with the 23andMe samples"
According to the CBC article the 23andMe results agreed to within 0.4%. The substantial differences in interpretation must be down to that 0.4% and/or the possibility that, as someone else posted, each analysis covers a subset of the genome. The 99.6% agreement would then apply to the common sections that were analysed. The differences between the variability of the 23andMe interpretations and the consistency of, say, the Ancestry variations are probably telling us something about the differences in the nature of the combinations of the interpretive algorithms and their underlying data.
Sunday 20th January 2019 22:25 GMT Greencat
23andMe's analysis/algos change over time
The first time I looked at my 23andMe ancestry test results, it showed I had a tiny percentage of native American ancestry. About a year later, their techniques had clearly changed as any evidence of said ancestry had disappeared from their report. I now currently have small amounts of Scandinavian ancestry being reported which were there last time I looked at my report.
Monday 21st January 2019 12:33 GMT Spazturtle
Re: 23andMe's analysis/algos change over time
You can't take the DNA of native americans and build a profile in the US due to the tribes lobbying the government. So when you see 'native american' on these DNA sites they have actually used the DNA of south americans for comparison, south americans usually have some european DNA, which causes people with some European genes to be falsely reported as being part native american.
Monday 21st January 2019 06:59 GMT Milton
I don't understand the confected debate over whether the sisters are identical (genetically so). The article explicitly states that they are, and—
"Boffins at Yale university, having studied the women's raw DNA data, said all the numbers should have been dead on."
—should leave no room for doubt. We may assume that the scientists were looking at properly collected DNA samples in a professional lab environment. Against that, some Reg readers' assertions about what they think they see in a photo displayed on a screen don't count for anything. The statement about height is particularly daft: nutrition, childhood illness and predilection for exercise cause significant changes to body form, and anyway ... maybe one of them was wearing heels for the photo?
As for why the results differ so much, one wonders about contamination. If the collection method is really as coarse as "spit in a bottle", why are we surprised there is DNA contamination in the samples? Sister A had a bacon sandwich; B had just washed her mouth with Listerine. (Or whatever. I won't be explicitly indelicate, but there are ways in which a mouth could become very confusing as a source for casual genetic sampling.)
I assume a much less "contaminable" process would mean taking a blood sample, or perhaps a speck of subdermal tissue, where other DNA cannot be? Your mouth is teeming with all sorts of junk ....
Monday 21st January 2019 09:07 GMT Paul D Smyth
"Despite having virtually identical DNA"
*Virtually* identical DNA. This is the great myth, even identical twins will not necessarily inherit the exact some amount of DNA from each parent so there is no guarantee that they would get exactly the same results. You would not expect them to be wildly different though.
As for the differences in results, anyone who has submitted their data to multiple companies will tell you that they get very different results, and that's with the same DNA test let alone different ones. This is because each company is using it's own dataset, i.e it's own set of DNA results therefore if they haven't got as many people for a particular region you will not get as many matches from that region hence the discrepancy. Now when you add the possibility that different tests will potentially have different markers recorded the differences will become even larger.
To sum up, anyone who expects a DNA test costing £79 to be as detailed as a DNA test costing £1000s is expecting a bit much. Some companies offer much more detailed tests but again, it's only as good as the dataset behind it.
Monday 21st January 2019 15:29 GMT iromko
Don't listen to naysayers
Reading this thread is like reading a flat-Earth society meeting notes.
Genealogy DNA testing has provided amazing tools for genealogists researching their roots and helped bring down many brick walls in their research. These tools are precise and scientifically sound. How else would they have assisted in finding lost relatives and biological families for adoptees?
Unfortunately, to make quick $$$, some companies promoted these tests heavily as a way to get your "ethnicity". Which is neither precise nor very scientific, but potential market for such tools is bigger. Efforts for proper promotion of really useful aspects of consumer DNA testing are left to volunteer bloggers, such like this:
Monday 21st January 2019 18:33 GMT David Beck
Re: Don't listen to naysayers
Thank you, saved me writing the same comment.
The DNA tests are fine, the interpreting for "ethnicity" is dodgy and different for each company as their databases differ.
What is not in question is the ability to match DNA for families whose members are tested. In what amounts to thousands of blind trials family members who did not know their cousin/brother/sister had the test got a resultant match when their results were returned.
Monday 21st January 2019 15:47 GMT Anonymous Coward
Slightly off topic
I knew a couple of local kids when I was young, called (by everyone) "the twins".
They looked very alike, and everything seemed unremarkable until their teacher, when they were 13(ish) asked "How can you be twins when your birthdates are a few weeks apart?"
They were a bit put out and explained that their dad told them that mum had become pregnant with one before the other was born.
At that point the teacher stopped asking questions and "the twins" then told us the story about the silly teacher and the explanation.
It raised a few eyebrows among the, mostly slightly older (15/16), group of friends, but we just let it lie.
It later transpired that dad had got two women pregnant in close succession and had worked with the mothers to create a foolproof plan to stop the kids from ever finding out.
Its a shame it was about 35 years ago. These days they could have had a night in a nice hotel and free trip to Jeremy Kyles studio.
Monday 21st January 2019 16:00 GMT Anonymous Coward
Triplets and DNA testing
In July 2917 an investigative reporter for Inside Edition looked into DNA testing for three identical triplets - https://www.insideedition.com/investigative/21784-how-reliable-are-home-dna-ancestry-tests-investigation-uses-triplets-to-find-out - and found similar differences... and there appeared to be significant differences between these identical triplets...Can only hope that official DNA testing is more rigorous and accurate. I see Parabon NanoLaabs is now doing DNA phenotyping which produces a description of the physical characteristics of the person's DNA...apparently law enforcement are starting to use it in investigations - with some success...https://snapshot.parabon-nanolabs.com/posters...Seems law enforcement are using it on old cold cases that have DNA evidence.
Monday 21st January 2019 19:03 GMT stiine
not read any coments yet....
But I think these two twins should wait about 6 months and re-run their experiment, comparing not only the new tests to the new test, but also to the old tests.
If possible, they should also try to get at least a dozen pairs of identical twins and submit them as well.
Some twins in the U.S. need to do this as well so that we can have DNS tests accuracy actually demonstrated to the courts.
Monday 21st January 2019 21:17 GMT AK565
But wait, there's more!
Given that most of the Mediterranean basin has been visiting each other for several thousand years, how exactly do these corpanies discern the differences? The same might be said of the Baltic. Hanseatic League anyone?
There's another issue: "Eastern European" is not one thing. The peoples living in that area belong to very different groups. They arrived at different times, speak very different languages, and had less mixing with each other than with their Germanic or Romance neighbors.