Should this become the official WikiTree policy on DNA Connections?

+23 votes
A recent G2G post brought to light the fact that DNA connections on a profile will "flow", even when there are "uncertain" relationships.   This led to a misunderstanding about "validity" of a DNA relationship.

I would like to propose that DNA connections do NOT populate onto the parents of a profile where the parents are shown as "Uncertain".  It is misleading to show a DNA connection when there is no "documented" connection to the parents.
in Policy and Style by Robin Lee G2G6 Pilot (544k points)
retagged by Ellen Smith
Sounds like a good idea to me.
But then Robin, how would it be possible to triangulate test results at deeper levels?
I guess it depends on how uncertain "uncertain" is. If it means "doubtful", it shouldn't propagate through at all. I tend to use uncertain a lot when someone has put a link to parents in and I'm pretty sure it's wrong, but I can't prove an alternative. Maybe it would just be better to have an option to not propagate, but I think on the whole I agree with Robin that if it's really an open question, it's a bad idea to propagate the DNA down the line. The biggest problem I've always had with DNA is people getting tests and giving me pedigrees with it that turned out to be very wrong. People often don't have very objective sense for how far back they can reliably go.
@Stephen, here is my issue....

one person puts on the web that the father of John Doe is Michael Doe (a known descendant of George Washington), with absolutely no sources or reasoning.   It is made up.    Now 15 people who are related to John Doe pick up this relationship to show they are related to George Washington, do you want them in your triangulation?
I want to start with the disclaimer that I'm still pretty clueless about DNA usage, although I am trying to get into this.

It seems to me that if DNA matches have been found that can triangulate, that should be able to prove the lines from all the tested individuals back to their MRCA.  If any of the people on these paths were uncertain, this should solidify them ... or what am I missing?
Decided this might more appropriately be posted as an answer than a comment....


You wisely said:

It seems to me that if DNA matches have been found that can triangulate, that should be able to prove the lines from all the tested individuals back to their MRCA.  If any of the people on these paths were uncertain, this should solidify them ... or what am I missing?

I fully agree, for the 'average' user on genealogy site, and the 'average' DNA tester, DNA is advertised and designed to help find genealogical connections - to prove lines and connection - not to be used to disconnect or disprove.  Not everyone has to be well-versed, or a data programmer, to get the basics, and keeping it simple for the average genealogy user on a user friendly genealogy site would be to keep it simple and for the DNA trail to pick up what many call the brick walls, where documentation is not available, ie., court records destroyed in fires, etc., and to solidify relationships by the DNA connections being visable.

I would think that would make sites like wikitree more attractive for users to become involved and get tested, and share their DNA more freely - when you pose limitations and make it complicated to fit your needs and not the average user - you lose everything that the genealogy site should be.

This is not suppose to be like "Pinky and the Brain" taking over the World here - it is suppose to be collaborative, open, user friendly, so that an accurate World tree can be created - by user involvement and the sharing their DNA to fill in the blanks.

But, Jacqueli,  we don't actually share DNA (test data) here; we only indicate that we have taken a particular test.

The actual comparing and connecting of our DNA data to other testers does not occur in WikiTree, but at the testing sites or sites that compare the test data of their membership.

WikiTree just shows the various paths a particular type of test will take in the tester's tree; that path will pass through the profiles of any individual placed in that path, regardless of the accuracy of that information.

That's my opinion, anyway - Narf!!

Robin, I feel that WikiTree's entire DNA system is flawed and should be scrapped until we can provide accurate guidance, as well as proper privacy protocols, on the subject.

edit: I believe that a couple of words used in comments below (placeholders by Steven Mix and non-biological by M Cole) sum up what I see as the flaws in WikiTree's DNA confirmation system.

Most of the WikiTree profiles we have created were created using non-biological information; thus, most of the profiles at Wikitree are merely placeholders waiting to be connected to the biological (genetic/DNA) information for each profiled individual. To connect the named individual in the placeholder to his/her biological information, we need his/her actual biological information and not a computer model of his/her presumed biological information.

Only with the actual biological information of each individual confirmed can we then confirm relationships between/among 2 or more profiles.

5 Answers

+18 votes
Best answer

I believe this may be a matter of semantics and improving the presentation of the information, not changing the underlying data or its intent.

The "DNA Connections" panel on the right-hand side of ancestral profiles is not a statement or even implication of biological relatedness or DNA matching. It begins clearly with: "It may be possible to confirm..."

I can't speak to the original design, but to me these are clearly to be used as hints, merely suggestions for further research. They are not reflections of DNA matches; they can't be because WikiTree stores no DNA data and cannot compare DNA data between one member and another.

The information in that panel is indicating only: "The people listed below posted on their own profiles that a DNA test was taken. Based upon the connections leading to and from this person on this branch of the family tree, there may be evidence to be had if the DNA information between/among these listed individuals is analyzed."

It's a research tool, not a statement of biological relatedness or condition. If we remove the tool's functionality, we remove a significant resource for WikiTreers to use in identifying, contacting, collaborating, and analyzing DNA information with potential cousins.

That said, it also seems evident--not just from this thread, but the many preceding it asking why a DNA "match" doesn't display on a particular ancestor's profile--that work needs to be done to make the purpose and function of that "DNA Connections" panel crystal clear and unambiguous.

by Edison Williams G2G6 Pilot (206k points)
selected by Peter Roberts

This is true...and while I understand the intent behind Robin's proposal (and perhaps in principle, even agree with it), Edison's statement here makes just as much, if not more sense to me, allowing people to now follow these clues to then use their DNA tests to prove or disprove these otherwise uncertain connections.

So, as Edison suggests, what needs to change here is the presentation of this data, not the removal of it.

This is an example where I think I would agree with Robin.

A living descendant of Louis XIV of France, and a descendant of his brother Philippe, Duc d'Orleans have the same Y dna R-Z381, which confirms that Louis and Philippe had the same father.

However this DNA information is now on every male ancestor of that line back to Charibert who was living about a 1000 years earlier in the  late sixth/early seventh century.  Not surprisingly Charibert and the following 6 or 7 generations are all marked as Uncertain given how little primary information exists in this time period.

Somehow saying that Charibert has the same ydna when there is no chance of proving that seems slightly ludicrous?

People aren't inclined to go looking for the comparisons that could break their connections and claims.  But the possibility should be pointed out.

Wherever a testee's own profile has somebody else's DNA of the same type flagged on it, there's a comparison waiting to be made.  Since this goes to the accuracy of the shared tree, I think there should be an expectation that this will be done, and a box to fill in to say what happened.

The connection between the testees can be found using the Relationship Finder.

But I don't see the point of flagging the DNA on all the dead people all the way back.  If you aren't a descendant, it doesn't tell you anything, and if you are, it doesn't tell you anything you don't already know.

I especially don't see the point of flagging dead ends.  You can't "confirm family relationships with Charibert" because we only show one line of descent - the earliest possible y-MRCA is centuries later.  And who cares about mtDNA on medieval men?  We can't dig up all the car parks.

Regarding the French royal line, the likeliest place for a break is two generations above Louis XIV and brother Philippe. Henri IV of Navarre was only a ninth cousin to the last of the Valois kings. The French went out that far because a couple of centuries ago they had resurrected an old tradition that inheritance could not pass through the female line, and made it officially law - the so-called "Salic Law". (They were trying to keep an English king off the French throne, so there was some point to it.)

Therefore, while we know the Bourbon Y-DNA "signature", the Capet-Valois "signature" is unconfirmed. (And Hugh Capet (c. 939 – 24 October 996) was elected king rather than inheriting the job - his connection to the previous, Carolingian dynasty is disputed, and only on the female side if it is valid.)

+15 votes
I find the Wikitree's support for uncertain connections (including the DNA Connections) as one of the most useful tools on wikitree.  Its in fact one of the reasons I've invested so much time here over the past year.  I feel like it really facilitates the kind of research that breaks through brick walls.

Do you have specific examples where this is creating a problem? In the process of checking DNA, at some point you'll check the relationship trail which does show the uncertain status.

It would probably be helpful to have DNA connections with an uncertain link to be labeled upfront, but fixing that might be more of an engineering lift than the team wants to invest in.
by M Cole G2G6 Mach 2 (23.6k points)
edited by M Cole

If you go back his Lee lineage, he has been VERY honest about "guesses" about heritage and parents, there are numerous unsourced profiles and lots of uncertainty in that lineage....yet his DNA goes all the way back and is now populating Lee Family lineages, including the Lees of Virginia with his DNA that does not match any of the known Lee DNA for those lineages.   It just makes the DNA connections seem meaningless.
"his DNA that does not match any of the known Lee DNA for those lineages."

Well, if I understand it correctly, then, when other male Lee's upload their DNA, his will be in the minority, and so it will become more clear where exactly to break the tree.

But all I see right now, going back to born 1670 is Tom's haplogroup R-U106. Unless there are others that are published in the bios somewhere, which differ.

I am not meaning to argue or be difficult. I am just wanting to get my head around the DNA business, because it continues to perplex me.
@Robin.  Thanks for the example.  I can see that especially since its the only Y-DNA test on there.  I could see myself looking at that and initially thinking that must be the correct haplogroup for that line.  (But I don't know that atDNA connection cause the same issue).

If the the connection were changed from Uncertain to non-biological, would that stop the DNA from populating past the point of uncertain?  If there is sufficient Y-DNA comparision for that line, then the relationship, if one exists all would have to be.  Although, I suppose a non-biolgical status implies a confirmed connection.  (Sorry just kind of thinking out loud).

Also, I'd say that if the DNA evidence shows that there isn't a paternal relationship, that's no longer uncertain, that's disproven, and the line should be disconnected.

I still think clearer labeling is the best approach.
You can't judge by majority vote, because it's quite easy for the majority of testees to be on the wrong side of the bad connection.  Some families proliferate, some don't.

In this case, the R-U106 line goes through a different immigrant and joins up at Coton Hall a few generations back.

But there's actually no real evidence that the Colonel descends from Coton Hall at all, so there's no real conflict yet.

There's just a lot of gaps in the genealogy and a war between people claiming their constitootional right to bridge the gaps with the handwaving of their choice, until proved otherwise.  Which doesn't work on a single tree, but it'll have to go to the Supreme Court.

What would really be fun is if one of the false descendants of the Colonel gets tested and turns out to be a match for the R-U106.  Then both R-U106 lines can be Confirmed as they stand.

But RJ, it seems to me that the DNA report is exactly serving its purpose.

The oldest profile that the DNA propagated to is Humphrey Lee (abt. 1506 - 1588)

Tom who uploaded the DNA report is allegedly descended through Humprhey's son Thomas Lee born 1535.

The R-U106 also propagates to Thomas' male siblings, and a certain number of male descendants, including Thoma's brother John Lee Esq.  born 1530.

There are several ways to then proceed. Presumed descendants of John can upload tests, to see if they are also R-U106. That would confirm with absolute certainty that both uploaders share the common ancestor, who has as a direct male descendant the *placeholder* profile currently displayed with the identity of Humphrey Lee.

Then also, somebody who thinks this is their lineage can also upload a test which shows a different DNA group.

And third, somebody can upload the same R-U106, but descended from an entirely diffferent Lee lineage.

So until we get each of those points of comparison, it is necessary to display the DNA results along the *presumed* path. That being said, I think WikiTree needs to have a better way to display these results, to indicate that the ancient profiles to which they are attached are merely *placeholders* for the *presumed* ancestor of the DNA tester.

So for example, now some male descendant who does not have any Uncertains in their line is encouraged to upload their own test. If it differs from R-U106, then it becomes possible to trace down where to break the tree on the R-U106 descendant line.

But it will also be necessary to bring in a third test, on either R-U106 or on the new uploaded test that differs, to work the triangulation. I think you will always need at least three points, which are somewhat widely spaced apart. In other words, distant cousins, who do not really know each other.

If you only get two close cousins reporting, those two basically just wash each other out in the data, as a single data point. So it is not a matter of counting the number of testers in any group. It is instead necessary to analyze their distance from each other..

Steven, it wasn't clear to me that you understood that U106 is a very broad category, estimated about 4 or 5 thousand years old, with over 100 million people currently.  Finding another tester that is also U106 will unfortunately not prove anything, as U106 males are everywhere in Europe and America.  U106 males are a part of R-M269, and probably most of the Lees with R-M269 are U106.  I'm U106, from Norway, possibly from Sweden or the Netherlands before that.

Unless you all take the Big Y test, comparing haplogroups is only good for rejecting matches, not confirming them, especially with haplogroups so high in the tree, so far from a recent SNP mutation.  Your best bet may be to test and compare as many STR markers as possible, the Y111 test with no more than a GD of 1.

I'd like to recommend reading this document:  U106 explored: its relationships, geography and history.  Page 8 has the history of U106.

Thank you Rob, that is very enlightening. And now DNA is even more perplexing to me.

Why in the world are the numbers so vastly different? And what in the world is the point of paying for and having any test result so immensely general as that R-U106? Why even bother to add such a thing to WikiTree? Might just as well say white, northwestern European.
Yeah, sorry!  Unless you spend a LOT of money, yDNA is only good for letting you know if you are on the right track or not.  If the major haplogroups are different, then you absolutely know you're on the wrong track, they can't be related on their paternal line (could still be related through other lines though).  If their haplogroups are the same, then you're on the right track, in the right crowd, but really can't prove anything beyond that.  You know that any further testing could expose wide differences.

yDNA STR testing can be helpful though, with as many markers as possible.  I personally think a minimum of 67 is needed, but 111 are better, provides a more reliable GD.  But a close match this way does NOT prove a path back to any specific ancestor, it only proves there is a path back to a common ancestor, but not who that is.  Paperwork / documentation still has to take you the rest of the way, to anyone specific.
+5 votes
I think we need to go half way:
If the relative has no DNA test of their own, show it as *possible* DNA results when the relationship is uncertain at any point.
If both relatives have DNA and the relationship is marked uncertain, it should come up for review to update the relationship status.
by Michael Babcock G2G1 (1.4k points)
+11 votes

The help page is a muddle, but from what is said about collaboration, cousin bait and DNA, it's clear that the Uncertain flag isn't just a passing comment on a connection that would be shown anyway.  It's an excuse for showing connections that otherwise wouldn't be shown.

And the major point of that is for the DNA to propagate, so that people will discover the matches that "prove" the uncertain connection.

This contradicts the principle that you can't confirm a trail you haven't got.  But it's what people think DNA is for.  They've seen the advertising.  They pay for DNA to find their genealogy, not to find fault with it. 


by RJ Horace G2G6 Pilot (492k points)

"They pay for DNA to find their genealogy, not to find fault with it."

With all the discussions I've had with people trying to work through non-paternal events, I think that's an exaggeration.

+4 votes

IMO, the current Wikitree Guidelines, as written and explained may violate the honor code.

Quality Assurance groups do not try a prove something works, they try and prove something fails.  In genealogy, DNA is used as a quality assurance Tool. IMO, In Genealogy DNA tests using only known trees are designed for the purpose of Quality Assurance. A positive match, using auDNA, between 2 Known DNA testers, even at the 4th cousin level, at least one segment that can be traced between those 2 DNA tests up to but not including the most recent common ancestors. 

Outside of Wikitree, a single segment means that there is virtually no chance of a Non-Paternal event because DNA, at least at certain levels, only returns the Positive Matches and Positive Matches don’t prove a negative. It is accurate to say that there is virtually no chance that both the documentation and DNA results will have produced a false positive. Wikitree disagrees, which is why we can’t confirm with DNA this part of the tree.

Outside of Wikitree, living people and DNA information are never public.  Whenever the results of DNA comparisons are published, a statement of a relationship between 2 DNA tests is accepted.  There is no requirement that a paper trail between these 2 DNA testers, even those customarily considered private, has to be public so that it can be independently verified.

My grandfather’s publicly available documentation contains the name of his Non-biological father.  7 DNA tests confirm a Non-Paternal Event as well as the identity of his Biological father.  I have explained this to all the living relatives affected by this event, and they agree.

But I will not provide all the documentation that allows for the public to independently verify relationships between two living people.  In my particular example, there is no documentation to verify independently.  It's an NPE.   This includes producing documentation such as birth certificate tied to a DNA sample, which would be necessary to separately verify a biological relationship when other traditionally accepted documentation tells a different story.

I am told that the honor code may be involved so I have removed the DNA confirmed status for those Profiles I manage, since (1) it can’t be used by NPE’s and Adoptees, and (2) I will not make the complete paper trail available to the public for those who I consider Private.

In my case, I named the biological father as the father, and the status as uncertain, because no documentation exists and the documentation that does exist is not publicly available. This is the closest I can get to being both accurate and stay within the Wikitree guidelines.

by Ken Sargent G2G6 Mach 5 (56.7k points)
But if your two auDNA testers knew they had 2 cousinships, they wouldn't know which to confirm.

Trouble is, they don't know they don't.

Related questions

+9 votes
4 answers
+15 votes
5 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright