DNA Confirmation - Updates on the App / What's New / What's Confusing

Question

This G2G post will likely be a longer than average read, but for those interested in DNA Confirmation, I would ask that you to take the time to read through it.  Thank you, in advance.

First off, as the author of the DNA Confirmation app that many use to add DNA confirmation source citations to their profiles, thank you for your continued support.  AND ... apologies for the long time in between updates.

The good news is that I am actively working on a series of updates (and have been for months), but, they are interrelated, and involved, and require LOTS of testing (in which I break a lot of things!) ... and I've got my hands in a few other pies, both in and out of WikiTree.  At the end of this post, I'll give you a sneak preview ...

BUT ... BEFORE THAT ... Let me just repeat a few principles about DNA Confirmation as we apply it here on WikiTree, and on which I have built the app - so we are ALL coming from the same place of understanding. (Help:DNA_Confirmation)

NOTE:  The points made below are in point form only, due to the limitation on the number of characters allowed in a G2G post. To read the more fulsome version, please visit this Free Space Page which gives much more detail.

(1) A DNA match is NOT a PROOF of a relationship.  In all but immediate family relationships, the amount of DNA one shares with another person could have multiple meanings.  If there is NO OTHER source to verify a relationship, a DNA Match cannot be used to prove anything.

(2) However, what a DNA Match CAN do is to CONFIRM a relationship that HAS been proven or supported by a valid genealogical source. 

(3) SO ... if you're with me so far, a DNA Match is not PROOF, but can be used to CONFIRM a relationship, that's great!  

The DNA Confirmation guidelines that we are working from on WikiTree at present indicate that for SIMPLE relationships, of 3rd cousins or closer, a DNA Confirmation can verify that type of relationship between any TWO people.  This is assuming - always - that there is a valid genealogically sourced trail to indicate this relationship.  

(4) MORE DISTANT ?  Well ... if your relationship is further than 3rd cousin (3rd cousins once removed or more, 4th cousin or more), then the amount you share will be relatively small ''(more about that later)'' - and - open to so many possible relationships.  To truly CONFIRM a relationship that distant, the standard is to find matches that can demonstrate that you share actual segments of DNA with on a common chromosome.  And, for a match this distant, you need 3 to TRIANGULATE your DNA Match, just to be sure.  

• This is the current policy, which has been in place for a number of years, and has been debated in other places on G2G, but is the foundation with which we are working (and my app is based on).  Some people suggest this is in fact too loose, others too stringent, which suggests it's about right, since it upsets people from both sides!  
• The current restriction in my DNA confirmation app, about only allowing 3 people for a triangulation, is one of the new features I'm building into the next version ... the newer version will allow you to add a 4th or 5th person to the mix after the initial three are triangulated

(5) TRICKY TRIANGULATION:  OK ... this is where it gets more difficult to apply.  BECAUSE of the way that DNA is passed on randomly from parent to child, except for the direct mother/father to son/daughter, where you KNOW you have 50% of each parent's DNA, all other ancestral relationships are variable.  Some ancestors, will have randomly gifted you with larger than expected amounts, and some of your ancestors may get shut out completely from your personal DNA lottery.

To validly compare three (or more) DNA descendants of a Common Ancestor (or couple), they must all come from a different Child of that Common Ancestor - or - possibly a Grandchild.  The analogy of a 3-legged stool is often used - if one leg is too short, or too long compared to the others, the stool tips over, as does our Confirmation evidence. (Think of looking at the tool from the top, where it connects to the common ancestor(s).)

* Corollary:   If an ancestor only has one child, you cannot do a DNA Confirmation by Triangulation to that ancestor, only to their child.

(6) UNIQUENESS IS KEY:  Remember that the purpose of the DNA Confirmation is to CONFIRM that which is already verified with a genealogical paper trail.  HOWEVER, that Confirmation has to be UNIQUE - you have to be able to say, that persons A, B, and C all descend from ancestor(s) X (and Y), and that they overlap on this ONE SEGMENT of DNA on THIS SPECIFIC Chromosome - and that there is NO OTHER WAY this could happen, because they share no other ancestors in common. 

• Oh oh ... Aye, there's the rub ... NO OTHER WAY ... that phrase is important, because if there are multiple ancestors that test takers A, B, and C share, then there is no way of KNOWING that THIS segment belongs to ancestors X and Y ... could belong to ancestors M and N ... or Z ???

* Corollary 1: This also implies that you have a fleshed out tree of ancestors for all DNA test takers at this ancestor's level, so that you know the common ancestor(s) is/are the only ones possible

* Corollary 2: If you have multiple relationships at different levels (a 4th cousin triangulation vs a 6th cousin triangulation), the closer one would always be the one you focus on.

SO.. in the case of endogamy, or pedigree collapse, where people are related to others in multiple ways, you CANNOT do a DNA confirmation IF there is more than one possible route from DNA tester to common ancestors , or if there are multiple ancestors to choose from. 

HOWEVER ... IF adding a 4th or a 5th DNA Test Taker into the mix can then further narrow down the Common Ancestor(s), and eliminate other possibilities - so that there is only one that remains - THEN we could in fact use that combo to confirm with DNA.  That is why I'm adding that new option into the next upgrade of the DNA Confirmation app.

(7) THERE ARE LIMITS .... and in the case of Triangulating DNA, specifically autosomal DNA, the type you get with most standard DNA tests, that DNA can only take you back around six generations, more or less.  That would take you to 4th great grandparents in common, that you share with 5th cousins.  POSSIBLY 6th cousins ... but ... certainly not much more.  If the shared DNA is too small it’s impossible to tell it from statistical accidents, and thus, it's unreliable to use as proof. The current lower limit is 7 cM, and in endogamous populations, it should be higher. 

(8) TO DREAM THE IMPOSSIBLE DREAM… it would be so nice to use DNA confirmation to take you back 10 generations, but our technology CANNOT do that yet. If you are looking at your pure paternal line, or direct maternal line, then yes, a Y DNA or mitochondrial DNA test match CAN be used to confirm those connections back that far.  But any other path, via autosomal DNA, sorry.

(9) A FINAL MEA CULPA. There are some deficiencies in the current version of the app. My apologies to those who have been led astray by those failings. Specifically, if there are multiple relationships between test takers, the current version of the app doesn’t detect or warn about that.  In fact, the algorithm is set to the find the CLOSEST MOST RECENT COMMON ANCESTOR for Test Taker # 1, and once it finds that, it goes from there.  It does not look to see if there are other common ancestors and warn you about that or advise on what to do (YET ... coming soon ...).  If two (or more) test takers do have multiple relationships, then reversing the order in which you enter them in the app may give you different results (and it should give you a warning, but doesn't, yet ...).   That SHOULD be a red flag ... which it isn't yet, but soon will be.

WHAT’S NEW ?

Well, here are some things that I do have working that are new in the development version of the app:

• Add DNApainter links to compare relationship predictions and cM values
• Add option to link to Relationship Finder (between DNA test takers) in the source citations (Turned ON by default)
• Add option to embed Compare link to GEDmatch, when both IDs defined. (Turned ON by default)
• Autosaves last used options for Cousin format, and new link options (via a cookie in browser).
• Revised drawing of MiniTrees from Simple Confirmation.
• Search for multiple relationship paths in Simple Confirmation.

WHAT’S NEXT ?

Before publishing this current update of the DNA Confirmation app, I want to finish testing all the scenarios in the Simple Confirmation (3rd cousin or less), including how to properly deal with multiple relationships, and proper wording for advice / next steps in those cases.

Once that testing is complete (soon I hope), then I will update the app to include these changes, and move onto the even larger updates:

• Add a check to confirm that parental relationships marked "uncertain" are flagged and that double-check that those marked "non-biological" are not used for DNA confirmation
• Add a Bio-Check (thanks Kay!) to check profiles, and warn if some in the chain are unsourced.
• Adding search for multiple relationships to the DNA Triangulation section
• Ability to add a 4th and a 5th DNA test taker to further refine the common ancestor
• Similar functionality added to X Chromosome matches
• Add Big Y test as an option for Y DNA matches

Anything I've left out?  As always - please do keep me informed of bugs you've found, or things that don't quite seem right (or in the new features especially, any wording that seems awkward or confusing)

FINALLY .. I must give a huge shout-out to the WikiTreers who volunteer their time to the time-consuming and often thankless job of checking recently added to profiles.  They do this to ensure the health of the tree, and are there to help.  Thanks folks!

:(Help:DNA_Confirmation)
 

P.S.  Final thanks  to Chris Whitten, Peter Roberts and Mags Gaulden who helped me with suggestions to ensure the accuracy of this post, that it was in line with the current policy, and reflects the latest thinking on the topic.

Comments

Thanks for that excellent write-up, Greg, and for all your work on the app.

Since you noted adding a link to DNA Painter, there's another you might want to consider. Only infrequently in triangulations do we see a specifically defined segment be the "least common denominator." Often, multiple larger segments overlap and form the smaller, in-common segment that is the one the triangulation should be working with; i.e., the segment demarcated by the highest in-common starting position value and the lowest ending position.

Granted that the positions (loci) are never entirely accurate because our inexpensive autosomal tests simply can't tell where a matching segment physically begins and ends, but a way is needed to get a calculation of the centiMorgan value for that in-common segment. The cM calculation is highly variable across different areas of the chromosomes and it isn't viable to just guess at the cM value of that in-common segment, especially if we're down in small segment territory.

When folks are doing one-to-one comparisons where no other tool is available to specify that cM count, the easiest solution I've found is an online calculator by Dr. Amy Williams: https://hapi-dna.org/query-cm/. It uses the GRCh37 genome reference model that all the major testing companies still use for autosomal DNA, so it's a simple entry for the in-common triangulated segment of the chromosome number and the determined start and end positions. Press the big blue button and you have your calculation.

A caveat is that Amy created the resource while running the Williams Lab at Cornell University, and she has since moved on from academia into a corporate role. But all indications are that she plans to leave that website in place.

Seems like it might be a useful adjunct within the app. I don't believe there's an available API (and the similar calculator at Rutgers University is very seldom online), but opening Amy's utility in a new window may be handy.

---

Thanks for that link, Edison.  Currently (in my development version of the app), I have the DNApainter link in the Simple Confirmation (just 2 testers, 3rd cousins or closer) after you enter the # of cM they share.   Sounds like this link you suggest would be most useful during Triangulation matches, where actual chromosomes and segments are compared.

 - Greg

---

Correct; for detailed comparisons within triangulation groups.

And...I didn't think your original post was at all lengthy. 

Answer 1

Thanks Greg, these sound like some helpful and interesting updates to your app.

For triangulations, would it be possible to improve support in the app for DNA matches who are not on WikiTree? Currently we can enter only one WikiTree ID of their closest relative who is on WikiTree. However, this limits the ancestor search to only that particular branch of your match's ancestors. It would be great if we had the option to enter say, 2 WikiTree IDs for the match's parents or 4 WikiTree IDs for the match's grandparents. This would help confirm that there is no other way for all the testers to be related (ie. no other common ancestors).

Comments

Interesting idea ...

Let me see what I can do ... i think I could accept either a single ancestor or an ancestral couple in this scenario, when looking for a closest existing WikiTree ID.  I hadn't considered that limiting it to a single ancestor you're only limiting the connection to half the person's tree.  I'll add that option to my to-do list now.

I think extending it to grandparents though would get messy (in terms of setup in the app for users, and logic inside the app for the poor programmer ...) - so - at that point, I think I would recommend trying to add more profiles to the tree, if possible.    I'll keep the idea in mind though, after all the other upgrades, and revisit it then.

---

Thanks, I realize that it might be complex to program/implement. It is pretty common for DNA matches to have living parents (or grandparents) so it is a tricky problem to work around when their profiles shouldn't be added to WikiTree. It isn't easy trying to work with the relationship finder to find common ancestors with a match's 2 parents/4 grandparents either. So that's why I was hoping your app could make that common ancestor hunt easier :)

Answer 2

Hello Greg,

thank you for your work on the DNA confirmation App.  I was wondering if you or others might have some thoughts on what might need to be included in a BigY DNA confirmation statement?

For an example, I have a 6C similarly BigY tested Y paternal line bloke who was located via paper research and asked to test, and who did so.  We have an authenticated paper trail with plenty of documentation (which is already on WikiTree), a 2/111 STR match, and a BY700 shared terminal SNP, and we have an average of two private SNPs.

I have constructed a preliminary confirmation statement outlining the people, their paternal line cousin relationship, the tests, the results including the STR 111 data, private SNP number, FTDNA kit numbers etc.  This also includes the % and date range information from the FTDNA STR MRCA calculator because of the current requirement to have this in the current STR match statements. There are already additional statements about autosomal tests and their Gedmatch kit IDs.

What else do I need to consider to document?

I have not yet even considered the even more remote, but paper trailable cousins also with matching BigY results.

Matt

Comments

My two cents is that the recently revised STR time to most recent common ancestor (TMRCA) information at FTDNA isn't of much use any longer. It's now a static table: the same data is presented based only on the calculated genetic distance (FTDNA uses the infinite allele model), and does not take into account the vastly different average mutation rates of the various STR values. The reference data I use ranges from the palindromic marker CDY which fairly rushes along at an average 0.03531 mutation rate, to DYS454 and DYS455 which, by comparison, are both a glacial 0.00016. A GD1 difference at CDY is not qualitatively equivalent to a GD1 at DYS454.

We get much more information from the Big Y reporting by looking at our list of matches, noting the non-matching variants, examining the Block Tree, and consulting the Discover^TM haplogroup report.

Not that the STR data can't contribute useful info, but it really can't pin down a TMRCA estimate. As one example, in our Williams DNA subproject we have, at 111 markers, men who are GD5 and for whom we've established a coalescence date circa 1690. Conversely, we have several GD5 matches for whom we know--from the SNP analysis--could not have had a common ancestor until circa 900 CE. Pretty close to apples and oranges but, judging only by the reported STR genetic distance, they should be approximately equivalent.

It would take some planning to craft the citation, but where Big Y results are available I very much believe WikiTree should use those rather than evaluated STR GD and the now static reporting. The reliance on STR comparison only is, quite literally, over a decade behind the times.

---

OK,

I understand why the 111 STR data might not be as useful as it once was when there was no BigY test.  The static nature of the comparison table is unfortunate, as is the lack of consideration for the vastly different mutation rates of different STRs.

What then should go into a BigY-700 DNA confirmation statement?  I am on top of shared SNPs, non-matching variants, private SNPs, Block tree and the test taker's neighbours (or lack of them), Y chromosome test coverage regions, and read numbers at individual locations, and the Disco haplogroup report.

I am starting though with a relatively straightforward R1b-DF27 example.  In this example there are actually 4 blokes, but I am only comparing 2 first off!  I already know how they all connnect on paper, plus a fair bit of information going back several more generations from the MRC Y Ancestor, and then back down to the present again from these earlier male ancestors.  More blokes to ask to test!

---

Thanks for starting this conversation, Matt, and carrying it on further in a separate G2G post.  I hadn't finalized the format I was planning on using (what new to add for the BigY versus what is used for a 111 marker match), since I haven't got to that piece in my app upgrade yet ... but definitely terminal SNP was going to be part of that.  There are some interesting suggestions coming in on your G2G post - so - that will help me out, and I will consult with my DNA team colleagues to come up with a solution.  Thanks for getting the ball rolling!