What do you think of our DNA confirmation instructions and possible automation?

+47 votes
3.7k views

Hi WikiTreers,

This message is aimed especially at DNA Project members and others with extensive experience with both DNA and WikiTree. I'd also be interested to hear input from members who have recently tried to follow our DNA Confirmation instructions.

I have been working on making the instructions simpler. My recent changes to Help:DNA_Confirmation and related pages are not final and need to be discussed by the community here.

The goal, of course, is to make the instructions easier to follow so that it's easier for members to do confirmation correctly. But I also think we are laying a foundation for future automation. At some point we could replace the need to manually mark relationships as Confirmed with DNA and manually create the source citations to justify them. It could all be done automatically based on information you enter about your matches.

For close cousin matches between WikiTree members I think it could be pretty simple. We already store information about the DNA tests members have taken. We would probably just need the member to select one of their tests and a test from another member, then enter the predicted relationship given to them by the test company or comparison site. If the members' relationship on WikiTree corresponds to the DNA-predicted relationship (and it's third cousins or closer) we could mark the appropriate relationships as DNA-confirmed, right? We could do something similar with yDNA and maybe mtDNA.

It would get more complicated if the match is not on WikiTree and/or we want to allow for triangulation. And I do think we want to allow for both. Otherwise the utility of the new system would be limited. If you can use the system to track all your significant DNA matches, I think more people would. I would.

But before I get too far into this, I want to get feedback on the changes to the DNA Confirmation page. What do you think of it? Any questions or complaints?

There are two things I would rather not rehash here.

First, the use of the word "confirmed." Like any word, its meaning can be debated. Its meaning on WikiTree is explained on the Confirmed with DNA page that's linked from the  icon.

Second, whether we should require public verification. In the past, we have considered and sometimes specified that DNA-confirmed conclusions should be verifiable on Y Search, Mitosearch, GEDmatch, etc. But I believe that building our requirements on top of sites like these isn't a solid foundation. We can and will work on building our relationships with other tools to facilitate comparisons (and there has been some recent progress on this) but our policies and tools shouldn't depend entirely on any one of them. Therefore, some sources that justify some DNA-confirmed conclusions will not be publicly verifiable. Educated genealogists will learn to put less stock into DNA sources that can't be verified, like with other sources.

Back to our confirmation instructions. If the DNA Confirmation page is agreeable, do you have thoughts on the Triangulation page? That has not been edited as significantly, but its content was controversial for a while. I don't know where people are at with it these days. (And I need to emphasize, I am not an experienced genetic genealogist. I haven't tried to mark any of my own ancestors as confirmed using triangulation.)

Triangulation is inevitably more complicated than one-to-one comparison, but we need to keep our instructions as simple as possible. What conclusions can users confidently draw from what they're getting from AncestryDNA, MyHeritage, etc? If we can boil things down to the barest essentials that we are comfortable with as a community, we can think about automation.

Another tangent: I am still unsure if we might want to store start and stop points for segment matches. This would enable the DNA conclusion to be automated, rather than entered. I think. But maybe we'd just want to depend on the conclusion given to the user by the testing company or third party. It's tempting to do it ourselves, but complicated. (We actually came close to starting to track segments last year, but we pulled back because of the GDPR and the privacy implications. Our hope was to create chromosome maps for ancestors. But we could not make these segment matches public in any way. Maintaining them privately is less beneficial to WikiTree's mission, so I have been inclined to leave it to third parties. But it's still tempting to think about it.)

We should probably consider automating simple one-to-one confirmation first. Then expand on that. I keep getting ahead of myself.

The question for now is whether there is input on the help pages. What are your thoughts?

Thanks!

Onward and upward,

Chris

in The Tree House by Chris Whitten G2G Astronaut (1.5m points)
The phrase "Corroborated with DNA" is probably a more accurate description of how we're using the DNA tests here at WikiTree.
Erik, He didn't want to rehash that.
Rob, fair enough.  Consider me 100% onboard with the current terminology.

well I read through the page again and if I'm following correctly I can mark my father as confirmed by DNA based on my brother's Y DNA test, so I did that. Previously I had thought that could only be posted on my brother's profile as the tester. My Mom and I are tested au DNA but my brother has not tested his au DNA. The visual does help.

The triangulation instructions seem fine to me, it does take some time to do those triangulation statements. I've only added to one line where there are large segments showing. But I was able to do a statement for my Mom's relationship to these 2 cousins also, she has the same as my relationship to these 2 cousins, but picks up an addional chromosome for her separate statement. To me it's important to be able to identify the DNA associated with a particular surname.

On another note, there's been a lot of discussion about reducing values, and to chime in there, here's an example, removing names, but all are wikitree members..... 7th cousins so of course tiny snippets shared with my kit T474191:

A915220, 8, 139,185,487, 140,339,538, 3.6, 309 and A042389 , 8, 139,373,435, 140,513,557, 3.6, 343.

A915220, 18, 6,631,427, 7,845,868, 6.0, 330 and A371231, 18, 6,959,761, 8,183,657, 5.8, 297

Well Chris, I took a look at everything and decided to try to confirm dna with my parents thru my sister's dna match.  I am not well versed in entering anything below the Biography line.  I looked at the instructions, but it still took me a couple of hours to figure it out. I would love to confirm a lot of my cousins/direct line but I feel it is very time consuming.  I wish there could be a way to enter your dna and have the system automatically confirm thru the relatives on the tree.
Maybe a link to another page with guidance re more unusual relationships would be helpful? What I needed guidance on myself was where the match is double fourth cousin once removed (predicted unsurprisingly as 'second or third cousin') ... i.e. myself and match have a link to 4 of my 4xgreat grandparents (so, 4 of her 3xgreat grandparents). Not an issue for me, but I know others where for instance match is with half second cousin, and not clear perhaps on that either.
I think segment identification on an ancestors page would be very helpful.
1. "I would love to confirm a lot of my cousins/direct line but I feel it is very time consuming."  What is time consuming is the format that must be used.  As long as all the elements are there, such as with any other source citation, what Wikitree problem could possibly be caused by the style used?

2. As far as adding GEDmatch account numbers?  That requirement begs lawsuits from people not even born yet.

3. The testing and comparison companies 'should' be producing source citations, just like familysearch.org does.  An example:  FTDNA match at 7cMs or greater between 'this person' and 'that person,' with 'this many' shared cMs and a predicted relationship of 'this.'  Then just the 'this person' and 'that person' would be changed to the Wikitree identification.  Another example:  GEDmatch between 'this person' and 'that person' at 7cMs and greater, with 'this many' shared cMs and a predicted length to MCRA at 'this number."

ok...back to my own.

B.
Hello B.,

How does adding GEDmatch account numbers beg lawsuits?

Thanks and sincerely, Peter
I'll have to give a private answer, so as not to encourage those who are character-challenged.

After (after) I posted here, I got an email telling of how 'my location' via my IP address must be given to GEDmatch so they can verify that I'm not a part of the EU and their new privacy law.  Until I do that, my GEDmatch information has been ported-out of their database and will not be available.

All they really had to do was show me a spot where I could opt-in for moving my data to Verogen.

So the public answer to your question is this:  (a paraphrase) Don't give the Chr #, start, and end locations on Wikitree.  I'll call that the 'folder contents.'  Just give the Gedmatch #, or, the key to the whole folder.

There are all kinds of responses on the internet following a search for "Should you give people your GEDmatch #."
Oops. That was me: Britain-141.

Sherrie Mitchell, you wrote:

On another note, there's been a lot of discussion about reducing values, and to chime in there, here's an example, removing names, but all are wikitree members..... 7th cousins so of course tiny snippets shared with my kit T474191:

A915220, 8, 139,185,487, 140,339,538, 3.6, 309 and A042389 , 8, 139,373,435, 140,513,557, 3.6, 343.

A915220, 18, 6,631,427, 7,845,868, 6.0, 330 and A371231, 18, 6,959,761, 8,183,657, 5.8, 297

A 3.6 cM isn't a valid segment size for DNA triangulation to a 7th cousin.

That's the problem when the tools are given but the users apply them wrongly. As per Itsik Pe'er presentation that I've linked several times here before (see source at the bottom of this post), a 3 - 4 centiMorgan triangulating segments (like in your case) is coming from up to 300 - 799 CE

That's 1220 - 1720 years ago, so a lot more generations. Hope you don't mind me pointing this out, it's not for the purpose of putting any blame on you or targeting you but rather a good example how quickly DNA confirmation is done wrong with GEDmatch which gives you the freedom to do anything with their parameters and results.

While I agree with some of Chris views, I disagree that giving the user the autonomy to enter any values (like in the example above from Sherrie) isn't helping WikiTree in any way to become a reputable website for DNA confirmations for relationships that require DNA triangulation.

I can guarantee that the majority of DNA confirmations requiring DNA triangulation entered manually by users will be wrong as they don't have enough knowledge about genetic genealogy.

Source:

Genome of the Netherlands Consortium. Whole-genome sequence variation, population structure and demographic history of the Dutch population. Nat Genet. 2014 Aug;46(8):818-25. doi: 10.1038/ng.3021. Epub 2014 Jun 29. PMID: 24974849.

I do have a Doug in my line. And a Phyllis keeps popping up with him. My mother is a Johnson. Adopted to Thompson. She was adopted at 4 months old. My email is lorirussell516@gmail.com

34 Answers

+5 votes

Feedback on current draft language on the Help:Triangulation page:

Triangulation has been compared to a three-legged stool. The common ancestor or ancestral couple is the seat. The three lines to the cousins are the legs. The three legs must meet at the seat. If two of the three legs are shorter because two of the three matches are closer to each other than to their cousin, the seat must be lower. You can only confirm relationships back to where the closest of the three matches connect. [Is this correct? May need editing. -- Whitten-1 14:19, 25 March 2019 (UTC)]

“Is this correct?”  I’d suggest not.  Even though I’ve used the three-legged stool analogy, it falls short when the testers aren’t descended from different children of the MRCAs. As a theoretical matter, I don’t see why the relationships shouldn’t still be confirmed all the way from each test taker up to but not including the MRCAs.  It’s been my understanding that the test takers should represent three mostly different lineages, but after reviewing as many articles as I can find on the subject, I don’t actually see that concept explained.  Since the DNA confirmation process is a purely Wikitree invention, there is no external rulebook to follow.

Just throwing out for further discussion some possible language to consider:

Ideally, the three DNA test takers should be descended from different children of the MRCA(s).  If not, the shared first in-common ancestor of two of the test takers should be no farther than a grandchild of the MRCA(s) AND those two test takers should be no closer than third cousins.

The definition page for MRCA may need to be updated to explain who the MRCAs are in a triangulated group.

by Kerry Larson G2G6 Pilot (234k points)
edited by Kerry Larson
Forgive me if this comment is out of sequence in the thread.  I wanted to add input of the concept of marking as DNA confirmed "up to but not including the MRCAs".

I guess on my wish list would be another option for between "confident" and "DNA confirmed", that would be "DNA supported MRCA couple".  That could be used when the supporting tests are from descendants of three or more children of the couple.
+4 votes
There is a step easier than triangulation or confirmation. It would be nice to show how many cousins share an ancestor on Ancestry using thru-lines. Each son-daughter could have number with cm match. Since this would be matching for each descendant it would be a table. Well maybe not completely easy. I have several 4th and 5th level ancestors with more than 20 matches from 7cm to 70 cm, many in the 15cm range. This must mean something!

**26 matches, 6 lines, Shepardson, Daniel 1775 - 1856 Daniel is the third great grandfather of Sue (23 members). ThruLines suggests that you may be related to 26 DNA matches through Daniel Shepardson.(Jonathan, Prudence-4, Millie-4, Daniel-3, David-12, Eri-2) Cook, Prudence 1780 - 1858 Prudence is the third great grandmother of Sue (23 members).ThruLines suggests that you may be related to 26 DNA matches through Prudence Cook. (9cm, 8cm, 13cm, 15cm, 7cm, 18cm, 14cm, 21cm, 18cm, 12cm, 16cm, 14cm, 10cm, 19cm, 8cm, 8cm, 9cm, plus sister and 1st cousins)
by Sue Hall G2G6 Pilot (168k points)
edited by Sue Hall
+1 vote
I liked the new instructions enough to start confirming.  Then i took liberties with mtDNA confirmation.  I'll be updating that with evidence after peer review of my triangulation by Denis Beauregard of the FTDNA French Heritage Project.  

I would love a triangulation helper, but thus far my triangulation has been hampered by tribalism.  Ancestry wont compare segments, so A ~ B and B ~ C attempts to suggest A ~ C, which is not necessarily true.   FTDNA is better, and I am finding MyHeritage would be if I paid for yet another site.  I am using DNA painter and Genetic Affairs, but they do not talk to each other, cannot take segment data from Ancestry, etc.

On the other hand most of my non-Ancestry matches where I can see segments have no tree.  Grrrr

Side rant aside, it's a great idea if it can function with the limited cohesive input data it will be afforded.  Or if you deal with the privace implications and allow direct uploads.
by Jeff Andle G2G6 Mach 1 (12.0k points)
+4 votes
Sorry if I'm a bit "late to the party"!

Two problems come to mind, that I don't think have been addressed yet:

(1) STEP 5: "Does your relationship correspond with the prediction?"

If there's even a point to this step - if some sort of quality check is being attempted, for example - perhaps somebody can explain it to me. If that's the case, fine, but we definitely need a better way because "as is" it's a big problem. We'd be much better off if this step was simply eliminated until something that produces more good than harm is written.

Here's the problem. Something like half the DNA tests in the world are done by AncestryDNA, and their "prediction" is seriously flawed. Sure, it tells me my two brothers are "Immediate Family", my nephew is "Close Family", and my 1st cousins are "1st cousins". But that's around where the trouble starts.

For the "1st Cousins" category. it tells the user: "Possible range: 1st-2nd cousins" In reality, this category covers 680cM to 1450cM. You'd be lucky to find even a 1C1R (1st cousin, once removed) to exceed 680cM. There's no way a 2nd cousin (2C) will. So it's just wrong here, and it gets worse.

That doesn't create a problem for us  YET - so let's continue on. My 1st cousin's son is on my list, at 286cM. It tells me that the possible range is 2C to 3C, when he's actually a 1C1R. So their "prediction" on him is wrong too (and, no, there's no reason to question that's his true relation - it's within limits for what you should expect, if you ignore AncestryDNA's "predictions").

It gets really bad in the "3rd Cousins" category. At least 9 of my 16 matches in that category are either 2C or 2C1R, yet AncestryDNA tells me the "possible range" is 3C to 4C. That's a LOT of potentially useful DNA confirmation matches being sidelined by "STEP 5" (for no reason!).

But for our purposes, the "4th Cousin" and "Distant Cousin" categories are the WORST. The "possible range" for these start at 4C and 5C, respectively. The problem is that these categories are really just 20cM to 90cM and 6cM to 20cM. Even a 2C1R could fall in those categories!

For our DNA confirmation, 3rd cousins are "gold". If you can find one corresponding to each of your 8 great-grandparents, you're done, all the way out to your great-great grandparents! Time to look at triangulation!

**** But on AncestryDNA, MOST of your 3C matches are in the "4th Cousin" or "Distant Cousin" categories, where the "predicted relationship" does NOT include 3C!!! ****

I have only four 3C matches in the "3rd Cousin" category, but DOZENS of them in the lower two categories. It's well known that 3rd cousins can go all the way down to 0cm - when they're below 6cM you get no match at all, which is why you sometimes don't get a match for 3C.

Because of AncestryDNA's flaw in this regard, our STEP 5 in the Confirmation process would tell a member NOT to use most of their most useful matches!!!

(2) We need to specify what counts as "third cousins or closer".

Genetically, we share, on average, the same amount of DNA with a 3C, half-2C1R, 2C2R, half-1C3R, etc.

Is a 3C1R "third cousins or closer"? Some people might think so, but I'd say "no". I'd also say that a half-2C2R is NOT "third cousins or closer", and neither is 2C3R. They all share LESS DNA, on average than a 3C.

The point is, it's obvious enough when you're talking about 1C, 2C, 3C, 4C, but when you start throwing in "half" and "removed" it becomes somewhat ambiguous what "closer than" means.
by Living Stanley G2G6 Mach 9 (91.1k points)
The thing I am finding hard to get my head around I the citation examples you how. I have been running a YDNA project for some ten year now and I am getting confused with all the words. Agreed a citation I needed but it need to be simple, so that everybody can understand what it is saying.

Take a BMD entry in the records and a certificate from one of these - How do we record the citation? With as few word as possible joe married mary on at there respective parent is certificate/record entry reference.

Frank, I agree with you. 

I think, rather than rely on a testing company's relationship estimate, we could ask them to check a chart like this where a range of CMs is shown for a given relationship. For example, 3rd cousins can range from 0 to 217 CM, 74 being average. 

Thanks John,

But be aware that that chart has it's problems too!

The irony is that I get so much resistance in my criticism of both, even while the one (the chart) supports my criticism of the other (AncestryDNA "predicted relation").

The 3rd cousin (3C) relation is a good example. From the chart, you'd expect that a 3C match runs around 74cM, and can go as high as 217cM. But when I see what matches I have for my own AncestryDNA test, and the one for my brother - I've found 93 3C matches so far - it's a very different story.

My average value is 46.1cM (vs Blaine's 74cM), and the highest value is 131cM (vs 217cM). For my own test, I have 69 matches over 40cM, and know exactly how 56 of them (81%) are related. Only two of the "mystery matches" are above the supposed mean of 74cM for 3C (their values are 115cM and 110cM, the latter of which is almost certainly a 2C).

So even if I found more 3C relations among my matches, I couldn't possibly have a greater max than the 131cM I quoted, only the 115cM "mystery match" could possibly bring the average up (and by only about 1cM), and all the other matches that might also be a 3C are below Blaine's 74cM (the overwhelming majority of which are 40cM or below).

I have only 9 matches of unknown relation that could bring my own average (46.1cM) up, and most would barely make that average move at all. But there are about 1000 matches in the 15cM to 40cM range alone that could bring my average down a bit. There's just no way my 3C matches would average more than about 46cM, even if you could find out how every single one of my matches are related to me!

Now, if you look at the .pdf file Blaine offers - which reports on the SAME data set used to make that chart - you can find numbers that are a lot more realistic. You can find, for example, that 90% of the reported values for AncestryDNA were between 14cM and 146cM, the average being 53cM - a LOT different from what the table says, even though it's from the exact … same … data).

The averages for 3C given for FTDNA and GEDmatch in that data are 88cM and 69cM, (vs 56cM for AncestryDNA) so the testing company matters - but that is NOT reflected in the table, which is "one size fits all".

But notice that even Blaine's average for 3C with AncestryDNA (56cM) is WELL BELOW the 90cM that AncestryDNA requires for its "3rd Cousin" predicted relationship. Further that AncestryDNA does NOT give 3C as a "possible" predicted relation for the categories below that 90cM - it misses not only the AVERAGE 3C match, but plenty of above average ones as well.

So, yes, Blaine's data (despite it's problems, including how it's presented) nonetheless confirms that AncestryDNA is not only wrong on this, but OBVIOUSLY wrong. I can only imagine some sort of "marketing messaging" is behind it. AncestryDNA is obviously more about selling to non-genealogy types than it is about actually being helpful to genetic genealogy.

As to reasons why Blaine's data could be having problems, there's (1) people could be calculating the relations they report incorrectly (2) data entry could have typos and (3) the contributors have not been able to find nearly as many of the low-value 3C matches as I have. That's why I tend to look at the central 90% values for the specific testing company category, when I resort to using Blaine's data. I use the chart for "even in the most ridiculous case" numbers.
+3 votes
So now that 23andMe has their DNA triangulation tool going, they're also allowing you to "edit relationship", which is data that's being collected to enhance their predicted relationships, I think. But the issue is that once I've edited my relationship to say "first cousin, once removed" I can no longer see what relationship it predicted, and there are no keys on the site to easily tell you what your shared amount of DNA would usually correspond to. The language in the citation for 23andme is no longer effective for any relationships that the tester has 'edited' in 23andme, and the page might benefits from a key about shared cM and relationships since "predicted relationship" can essentially be erased.
by Jillian Kern G2G4 (4.1k points)
You can recover the original predicted relationship.  On the match page, under the "Your genetic relationship" section, select the "Edit Relationship" link.  At the top of the list, you can choose "Revert to predicted".  You can then use that information for Wikitree's DNA confirmation statement, and then you can re-edit the relationship.

"Revert to predicted" is only an option if you've actually edited the relationship.
+2 votes
I think it's important to use the data to confirm/validate evidenced-based associations. The most impactful validation of a genealogical relationship inferred from public or private data is an unequivocal match in DNA. Some tests are more meaningful than others - the ones that should be avoided are the tests that produce gray-area or inconclusive data. For instance, autosomal DNA tests are poor tests to validate genealogical relationships. So are Y-DNA STR tests - too error (Type II) prone. The more definitive tests are NGS/SNPs for Y-DNA  and Full-Sequence for MT-DNA. This would eliminate the need for segment comparisons, a minimum # of STRs, or autosomal scrambles to discern the strength of genetic connections on public trees.
by Leake Little G2G6 Mach 1 (16.2k points)

I've added my DNA matches to some of my Paternal direct line. i've kept it simple and in a table format. This link shows 11 of the DNA matches I have for MRCA William Gambrill (1821 - 1888). I have added my name and the initials of the DNA match, the cM, place tested and relationship - actual or predicted. I've never had the need to use predicted but added that anyway. I only claim a DNA match when I know how we relate. For the purposes of adding DNA info to a deceased persons profile, I dont think anything else is needed or necessary. Their is additional DNA information about me on the right side of the page. 

+2 votes

I've been thinking about these rules lately as I have been attempting some triangulations.  I posted some of these thoughts as part of another question, but they would be well-suited here as well.

Currently, there is a rather large discrepancy in the effort required to confirm the MRCAs of third cousins and those of third cousins once removed (and more distant).  I would be more inclined to define the "three-legged stool" to be more consistent with the one-to-one confirmation requirements.

Consider three DNA matches.  The three share "major" MRCAs, and two of them share a "minor", more recent MRCA.  The current stool rule allows the minor MRCA to be up to one generation below the major MRCAs.

I would propose that if we are allowing one-to-one DNA confirmation of third cousins, we should also allow triangulation in which a minor MRCA can be up to a third cousin of an ancestor in the more distantly related DNA match's line, where that ancestor is in the same generation of descent from the major MRCAs.  This would be the equivalent of allowing the minor MRCA to be up to 4 generations below the major MRCAs.

This seems like a lot of generations to allow.  However, if this concept seems as reasonable to others as it does to me, I might also propose tightening up the one-to-one requirements as well, maybe to second cousins, as third cousin triangulation requirements would remain fairly relaxed.  The biggest issue there is that it would require revisiting many pre-existing DNA confirmations.

by Brian Lamothe G2G6 Mach 4 (43.2k points)
edited by Brian Lamothe
I had also suggested changing the "three-legged stool" as a bad analogy in my answer to this post. There wasn't any feedback from someone who wants to defend the thre-legged stool, but nothing happened either. I still feel the analogy is bad -- the point of the requirement is having "independently sampled" segments from the three individuals. That would translate into keeping the stool legs from joining close to the ground, not keeping them from joining up close to the seat.
+3 votes

Going through this thread, I see lots of discussion about the proper place to put the DNA source citation, but what is the final "official" ruling?.

Personally I think it should go as a ===DNA=== subsection under ==Sources==, but the Help:DNA_Confirmation page is self-contradictory saying put it under ==Sources== but also maybe under ==Biography==>subsection ===DNA===?

by Joe Murray G2G6 Mach 7 (77.8k points)
edited by Joe Murray
I don't believe there is a final "official" ruling. If it makes sense to you to put a subsection under == Sources ==, then I would put it there.
Thanks, Kay.

But it would still be nice to have a consistent formatting decision across WikiTree so it's in the same place for every profile (at least going forward.)

Much of the other formatting and styles are set as a standard, so I don't know why Administration is being wishy washy about this item.
+2 votes

Some comments on the Triangulation page:

1. I think a search for endogamy should be added as a requirement. See (3) below.

2. The writing in the all-important first paragraph needs to be tightened up. Right now, it says:
"Simple one-to-one DNA confirmation is sufficient if your match is a third cousin or closer. Confirming relationships using more distant cousin matches requires triangulation. Three or more cousins need to all match each other on a single segment of DNA that is at least 7cM long. Seven cM is a bare minimum and presumes that the cousins' relationship back to their most recent common ancestor(s) is well-documented."

The triangulation process is so different than the usual, with so many new requirements, that people coming to this page need to be hit in the face with the main differences all at once. Sort of like the pre-1700 quiz -- there needs to be something significant to grab attention. (Would a triangulation certification be reasonable?) That first paragraph right now just sort of eases into it. I would cut the first sentence entirely, and suggest a revision something like:


Confirming relationships with matches more distant than third cousins requires triangulation. To begin the triangulation process, you need at least two different matches to your kit, both at the same DNA website, and those matches themselves cannot be closer third cousins to each other or to you.  Once you have identified these two other cousins, triangulation requires you all to match each other on a single segment of DNA that is at least 7cM long. To check this, you will need to go into a chromosome browser and see that the segments you share with each match are both on the same part of the same chromosome (details below). Right now, Ancestry DNA does not provide a chromosome browser, so triangulation cannot be performed if your matches are through Ancestry DNA. Details on how to triangulate at the other main DNA services are given below.

Besides having a triangulated DNA segment between all three matches, you must also have identified a common ancestor in all three of your trees. Your trees must then meet three requirements: 

  • your line and your matches' lines to the common ancestor must all be well-documented with original or otherwise reliable paper records. The profiles on WikiTree in your own line to the common ancestor must all include this documentation.
  • your three lines to the common ancestor must connect in a Three-Legged Stool (see below)
  • you must have conducted a thorough search for endogamy (see below)

3. Concerning the thorough search for endogamy, there are two issues that arise: (1) pedigree collapse, when one person has multiple lines connecting back to the same ancestors (i.e., the parents of some ancestor in the line were some sort of cousins to each other) (2), when one person and a match have two different pairs of common ancestors in their trees --i.e., Alice and Bob Smith and separately Carol and Daniel Jones. This scenario doesn't require that cousins have produced offspring.

In the second case, I think we should say that triangulation just won't work with this particular trio of cousins. In the first case, any connection back to the MRCA which could conceivably have been bypassed as the DNA was passed down to the testee should not be marked as confirmed. But other connections that could not have been bypassed could be marked as confirmed.

Endogamy can be a huge issue with triangulation, so I think we cannot avoid include this additional requirement and complication if we want to keep the confirmations on endogamous communities mostly correct. We would need to spell out the requirements for a "thorough search" and for how distant two different sets of ancestors would need to be to spoil the triangulation (i.e., if you and your cousin share a set of 4th grandparents and then separate lines going back to John Howland of the Mayflower, that second line is almost certainly not the source of the shared DNA and can be ignored). And there will probably be disagreement on what these requirements should be. But I imagine there is agreement that endogamy is a big issue with triangulation.

4. I continue to think the Three-Legged Stool is not currently setup correctly. It's intention is to ensure independence of the matches (i.e., if you share a segment with someone, you are suddenly much more likely to share a segment with that person's sibling -- that's dependance). But the current setup should be altered to truly get at what independance would meen. See my other answer to this question and comment on another answer.

by Barry Smith G2G6 Pilot (291k points)
edited by Barry Smith
+2 votes
I tried Triangulationand had to have help with the citation. Also ran into different tools giving different results as a match in visual triangulation did not match on one to one. Think many people would think as I did if You went in GedMatch from one to many to triangulation that is triangulated. On one to one tool though the segments were not as they appeared in visual triangulation and one kit had no overlap. I would love to see this automated. I just subscribed to DNA painter and while I have a lot to learn there the idea is to be able to assign a particular segment to a particular ancester and that would be awesome. I started by taking my Most Recent Ancestors from GeDmatch as I pretty well know where that DNA comes from. You can designate it as maternal or paternal and even designate the grandparents it came from.
by Sherry Holston G2G6 Mach 2 (22.5k points)
+2 votes
I have studied biology and genetics in grad school several years ago and have been working on  discovering my family through geneology for the last three years and must admit the instructions for confirming my own mother on here are either difficult for me or there is a glitch with my profile. We have been confirmed with a tag on our profiles since we created them and uploaded our dna through the various sites but I continuously have a "suggestion" to fix our status. I have followed the instructions and made changes on multiple occasions- even going so far as to copy and paste and example and filling in our names/numbers but the error won't go away. I haven't even begun triangulation as I feel it would not register if I can't get this one done and I am someone who likes to clear the suggestion page. If you have any ideas or could take a look I'd be grateful. I am in the process of changing my bio again and moving a === DNA === section under the Bio as it says some people have done and am going to change to the GEDmatch.com again vs Ancestry to see if this combination works. Cheers! Becky Elizabeth (Simmons-11603)
by Becky Simmons G2G6 Mach 2 (26.6k points)
Becky, The reason you're getting the "suggestion" is that it's looking for your confirmation to begin, "Maternal relationship is confirmed by ...." instead of "Mother and daughter relationship are confirmed...." Change that and the suggestion should go away.
+2 votes
See:

https://apps.wikitree.com/apps/clarke11007/DNAconf.php

 Greg Clark's DNA citation app.
by Loretta Corbin G2G6 Pilot (243k points)
+1 vote
The whole DNA section of WikiTree is poorly executed. There is too much reliance on outside primary data sources for these data to be verifiable within the WikiTree community. Stop trying to emulate genetic genealogy tools for which you cannot control the primary data! Chromosome painters, triangulation, etc. are tools for owners of these data who have primary access and the confidentiality to use them. No third party can enter that bubble presently - nor should they.

I recommend scrapping the whole "Confirmed by DNA" program and revert back to what data can be shared in a public forum with unaffiliated third-parties (distant cousins as well as genetic strangers). And those data would be simple - (1) Y-SNP terminal haplogroup for males; and (2) mt-SNP terminal haplogroup for females.

Users would have to be tested at this level for intergenerational tracking of DNA for 'validation' purposes. And that's the essential thing here - Chris and most of the Wiki team are focused on "quality assurance" for a more granular system in which no one on WikiTree can confirm the accuracy of the data inputted except the owner of the primary (DNA) database. I would rather focus on the "quality control" of outcomes using DNA data.

This means that a capital reference is used communicate what is essential for genealogy; that is, a relationship has been validated by outside DNA testing. In this case terminal haplogroups include all of the underlying STRs, autosomal data, etc. They are intergenerational and unique to a specific genetic line (family) and group naturally into genetic clusters that don't need to be confirmed by WikiTree users at the profile record level. WikiTree can simply maintain a central index of these data (updated as needed) and individual users can associate profiles with these control data. That eliminates the need for individual users to access the private test results of distant cousins (or tree-trimmers) to 'confirm' anything. The user then must only cite the DNA database where these results are held. Or one could cite other sources such as The Big Tree (ytree.net), etc.

These data cascade up and down as testing hypotheses for suspecting relations. No one wants to test ad naseum but painting chromosomes for distant relatives is not the purpose of genealogy per se. Autosomal data are fun but not definitive. Similarly STRs provide some direction for haplotype but do not confirm haplogroups uniquely. There are similar distinctions with mt-DNA data. SNPs are definitive for men and women.
by Leake Little G2G6 Mach 1 (16.2k points)
0 votes

Hi,

I just found DNA for my Grandpa. But don’t know how to add anything saying it confirmed. My kit number here B25630. I have other DNA love to get in DNA group on Discord too. I have so many questions. I do need help verifying statements and such on Wikitree if anyone could help it be appreciated. I just added link to DNA to profile as well. 

https://www.familytreedna.com/groups/bright/about/results

https://www.wikitree.com/wiki/Brecht-13

Billie

by Billie Keaffaber G2G6 Mach 4 (41.2k points)
As this question is old....3 years ago, I would suggest you post a brand new question asking for help.

Related questions

+12 votes
4 answers
+10 votes
2 answers
+9 votes
1 answer
+2 votes
1 answer
+4 votes
2 answers
+6 votes
2 answers
+9 votes
1 answer
+7 votes
2 answers
+4 votes
3 answers

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

disclaimer - terms - copyright

...