Proposing a rule change for DNA confirmation standards

+6 votes
712 views

So I know ThruLines is still in beta, but I think that with this new system it is reliable enough to expand the DNA confirmation standard.  Right now the requirements are 3rd cousin , or triangulation through GEDmatch.. but not many testers use GEDmatch.  So I'm going to propose a new rule for leadership to consider.  I'll also give a personal example of how there really is a DNA confirmation in my family tree but per the rules I can't add it to wikitree.

The Proposed Rule-

1- A confirmed primary source record (Census, Will, Estate Documentation, etc) showing  the MRCA's direct connection to his/her children.

2- Relationship match from 3 separate progeny lines of the MRCA, of 3 or more individuals, sharing at least 7cM in 1 segment.

The proposal is to jointly use the the new (easily accessible) sorting of ThruLines and primary sources.  To date there isn't a marriage of the 2 on wikitree and I think it's time for this to occur. 

My Example-

Here's an example of how/why this should be implemented-

MRCA- https://www.wikitree.com/wiki/Yount-761

The primary source listing sons and son-in-laws (sons George and John and son-in-law Amos McDaniel that married Catherine Yount for this example)- https://www.wikitree.com/photo.php/7/7b/Yount-761.png

ThruLines Connection-  There are too many generations between me and John to have any matches, but 2 other relatives Robert Mesnard and Cheryl Harrison from my branch gave me permission to research. (I already specifically asked for permission from both of these individuals to post/share their information).  Here are the results through ThruLines-

(I think the images were too big to add in here, so here are the links)

https://www.wikitree.com/photo.php/f/ff/McDaniel-4979-2.png

https://www.wikitree.com/photo.php/4/4e/McDaniel-4979.png

https://www.wikitree.com/photo.php/f/f0/McDaniel-4979-1.png

Looking at those links you can see that not only is there a "triangulation" (not data triangulation but a general 3-way match...using the term loosely) with bjcowan1947 (still wouldn't fit the current rule because no GEDmatch), but also there are good cM connections for both testers to relatives belonging to other progeny of John Yount.

Under this proposed Rule, Robert's test alone would not be enough because he does not have 3 progeny lines in his results.  Cheryl's results would be good enough alone because she does have 3 progeny lines listed (one is private but with Roberts you can see it is John Yount Jr).  Combined together Robert would have the match because with him & Cheryl (1 progeny), matching progeny from both George and John Jr (2 other progeny), it'd fit the rules.

While I personally think that a connection or "triangulation" (again using the term loosely) on Thrulines should be able to stand on its own, I think that having a primary source accompanying the testing with 3 progeny lines is more than enough to help expand the special DNA confirmation tag for profiles. 

Hoping others agree or will have comments/ideas, and i hope the leadership considers this and allows for further DNA confirmation options.

Thanks!

-Eric

in The Tree House by Eric McDaniel G2G6 Mach 3 (39.1k points)
edited by Eric McDaniel
You lost me in the first sentence. What's ThruLInes?
If you're familiar with ancestry DNA, they have an option for finding matches within a "circle.". Ancestry introduced a new update/change that is very nice and helps sort your DNA matches.  It also makes proposed ancestors based on others' trees.  So the reliability of a match still depends on the circumstances, but in certain cases it is a huge step forward for expanding dead ends.
I am an adoptee in reunion with both parents. There is also a recently discovered NPE in my maternal line. I have 25 ThruLunes to supposed ancestors from the NPE line to whom I am not actually genetically linked. Unless Ancestry makes major modifications to ThruLines, they are only as reliable as the trees posted on Ancestry.
Correct!  I've found the same issue of "claimed" lines that I know are false.  And that is why I think this system can be used to elevate relations to "confirmed with DNA" if there is a primary source to accompany/validate it.
Pam, ThruLines is flawed in that it depends on the accuracy of the trees it uses to connect good DNA matches, but that does not mean the matches are wrong though.  They appear to be good, and you can tell how good by the total cM AncestryDNA reports.  If you have 25 ThruLines to related matches, you are clearly genetically connected to them - the problem being *how*.

What ThruLines appears to be doing is to take good DNA matches and then examine yours and others' trees for ways to connect them.  If others have faulty trees, then AncestryDNA makes faulty connections.  What you need to do now is to figure out how you *are* connected to all of those.  25 of them going in a similar direction sounds like it could be a very good clue, to an unknown connecting ancestor!
Rob,

Actually, I'm not. The problem is that someone I am closely related to (birth mother) has the NPE listed in her extensive, well researched tree. Ancestry sees my DNA relationship to her and then picks up that paper, but not genetic line. I just edited my original comment from "25 individuals" to "25 supposed ancestors." Hopefully that will more clearly communicate what I mean. I didn't open every "potential ancestor" from the NPE line, but the couple I did all tie to my birth mother's or her half sister's family's trees. I am genetically related to each of them through my maternal grandmother only not the paper maternal grandfather.

Pam
I think that's a solid proposal that would work. I think the only thing I would add to it is that the amount of shared DNA should be within statistical norms for the putative relationship and the confirmation should be rejected if the amount for one of the matches is either way too low or way to high for what's expected. I've worked through more than 60 thru lines on ancestry now and I find that all but 8 fall within the expected percentiles. That's actually very encouraging and for me it says that the reported relationship is in fact the relationship.
ThruLines just showed up on my tree yesterday at ancestry.com and I'm not impressed at all.  It relies too much on family trees which may or may not be accurate.  An example illustrates:

My paternal grandfather was born in Germany and came to US. He told all 8 of his children that he was the son of Christian Gottfried Heinrich Friebel and he didn't know the name of his mother, but she died when he was about 2.5 years. This info had been passed down to us later generations. A few years back I got the church record microfilm of where he was born from familysearch.org.  I knew his DOB.  I found his baptism record and it turns out he was born illegitimate to J. Albertine Friebel, the sister of the man he said was his father.  No death record was found for Albertine, but it seems likely she died when he was 2.5 yrs. old and then he was raised by his uncle, thinking his uncle was his father.

I can find only one family tree at ancestry that still shows Christian Gottfried Heinrich as his father, yet ThruLines suggested him as the unknown father in my tree. It was based on this cousin's incorrect tree AND she hasn't even been DNA tested.  So why did DNA ThruLines come up with a recommendation?

I immediately turned off DNA ThruLines just like I've turned off family tree hints.
When I am confirming lines, I try and compare the lines until a couple lines from the living and present time. I leave the people that tested private.  Like 4th great grandparents going to their grandchildren's that tested. When we get enough lines on wikitree the ones that tested and add the DNA lines will be able to use the parents or grandparents as the DNA line.  If we all have 1,000 DNA cousins we should be able to go around privately to the lines that have deceased to confirm DNA.  The Trulines are helping me find my 4th or 5th great grandparents DNA lines quickly to DNA matches.  I am adding the lines to Wikitree while I decide how to confirm the lines privately.

A bit of empirical data: on his segmentology blog Jim Bartlett found ~10% of his ThroughLines were misleading:

https://segmentology.org/2019/07/29/a-few-thrulines-are-false-and-some-are-misleading/

Thrulines does not add anything that a confirmed trail of primary source records to the MRCA doesn't. All it does is aid in finding people with you you might be able to build such trails.

So your rule change seems to me not substantially different than allowing confirmation once there are paper trails down three progeny lines to three people who show up as pairwise matches, each on at least one 7cM segment. IMO this is far too weak. I have, on several occasions, identified triangles of pairwise matches where one of the matching segments turned out later to be provably not from the suggested common ancestor. Heck, given three 7-10cM segments, there is probably at least a 50% chance that one of them is a false match. It is true triangulation that *significantly* reduces the chance of a short segment being a false match.

8 Answers

+28 votes
 
Best answer
I have to point out that ThruLines cannot triangulate DNA, it can only show suggested tree connections and the "fact" that each person shares some DNA - somewhere

Ancestry does not provide the segment data necessary for triangulation.

All the other main testing companies do, and GEDmatch of course (MyHeritage, FamilyTreeDNA, 23andme)
by Lorna Henderson G2G6 Mach 2 (21.1k points)
selected by Maggie N.
True, my second comment on a "triangulation" I am using the term very loosely.  I only meant for the purpose of showing there are 3 individuals matching each other.
MyHeritage, yes

23andMe, yes

FTDNA, no

While FTDNA does have a chromosome browser and A can confirm that he matches B at the same place as he matches C, he cannot confirm that B matches C without contacting either B or C and have them possibly confirm that third edge of the triangle.
Thom
I didn't say that the other companies provide triangulation, but that they DID provide the segment data necessary that Ancestry refuses to provide.

Yes, you have to contact both parties on FTDNA to prove there really is triangulation of the segment being investigated - or find them on the other (non Ancestry) sites.
Ancestry ThruLines are primarily based upon Ancestry Family Trees. They are nearly useless for DNA Confirmation. I have ThruLines connecting me with ancestors which are wrong, but nonetheless listed in many of my cousins' family trees. In other instances, share DNA with cousins through other common ancestry, rather than that mistakenly suggested by ThruLines. Using ThruLines to confirm older ancestors would be a disasterous practice.
+14 votes
I have to counter this with option 2 for NPE's... NO document (who documents an affair?) and some larger number of intersecting matches to MRCA.  In my father's case, I have a half first cousin, a 2nd cousin once removed, and a slew of 3rd to 6th cousins.  I can use differentials between the degree of match of the two close cousins to tell which of their parents (father of one, grandfather of the other) is the correct grandfather for me.  the more distant cousins all intersect several generations earlier.  I did it manually before the new tool on Ancestry, but I thing 8-10 4th to 6th cousins are enough to call it confirmed!
by Jeff Andle G2G6 (9.5k points)
I agree that is a very tricky issue to try and map out.  I would say the weakest issue we have is the fact that we can't tell for sure if the cMs is from John or his Wife.. BUT I would say that if you have 3 progeny lines all matching I wouldn't think an affair has occurred (otherwise that's a long lasting affair!).  But lets say one did, a GEDmatch triangulation wouldn't necessarily catch it if you had 3 testers like my example.  I think the proposal wouldn't necessarily increase the risk of such circumstances compared to what is already available.  But I'm no DNA expert though...
3ggf and ggm intersect all matches in question; through a few paths

2ggf intersect most through a few paths

ggf intersects 2 through two sons

gf intersects one directly and is uncle of other

gf to f was the npe.  it was not long term.
+10 votes
I have at least a couple ThruLines where several cousins do share a common ancestor but most of them have the wrong one in their trees, copied from a bad family tree. This keeps putting that incorrect ancestor as a potential match for me. I know it’s wrong but other genetic cousins might not and also add that bad ancestor, making the problem worse.

Other than a better paper trail, my DNA proof that my ancestor is correct is many DNA matches upstream as well but ThruLines ignores that. ThruLines needs to be the start of an investigation, not the end of one.

Ultimately, DNA is only one more piece of evidence and sources do not equal proof, even triangulation is not proof. What, exactly, does the Confirmed with DNA supposed to indicate then?
by Davis Simpson G2G6 Mach 2 (21.5k points)
I believe the extra designation here on wikitree helps prevent incorrect removal of relationships and to also help others know that the relationship has been thoroughly investigated.  That's the main benefit in my mind, but I'm sure others have their reasons for wanting it.
+8 votes
So I am guessing that these "thrulines" for Ancestry, are either using the Autocluster tool OR doing something similar to the "Theory of Family Relativity" that My Heritages uses. Both of which are fairly new tools that were introduced last week during Rootstech.

As someone commented above - the connections are only as accurate as the trees that have been added, and in my family, I have found some trees to be just plain WRONG. I means the person that makes the connection between me and the DNA match may be right but they are not married to or related to the person on MY tree.

As for the Autoclusters, I found them useful for knowing who else on my trees, the new DNA matches are related to, but without a tree - and many of them have NO TREE - I cannot even begin to start looking.
by Robynne Lozier G2G6 Pilot (722k points)
Yes, a few people have pointed out that it can't be taken at face value.  That's why I would propose an expansion only if there is primary documentation to help confirm the connection.  Outside of lacking sources, most of my other false matches are when all testers are from the same progeny.

"primary documentation to help confirm the connection"

That's the problem. On WikiTree DNA can be used to further confirm the paper trail, never the other way around. I expect in time that philosophy will be forced to change, but probably not real soon. The current system makes it almost impossible to recognize ancestry through an NPE.

I have just found the biological father of a relative. Since there is no paper trail, I will not be able to connect my relative to his biological ancestors, but WikiTree's rules will let me connect him to the ancestors of the loving but non-biological father he grew up with. 

+6 votes
I am hoping the people that tested DNA will have a DNA tag and a program or tags to highlight them as they are on trees that they have DNA matches through a line. With Thrulines if you have an ancestor with 29 DNA matches, with 60 DNA matches and you go through each of those lines adding them on wikitree to the last relative not living we should be able to get a DNA picture, an overview that if there are that many DNA matches maybe more people that DNA tested will overlap and confirm that they have some DNA matches to that line also.  I was taking a look at my dads 5th great grandpa Deeter-44 and he has 79 DNA matches the ranges are 17cM 8cM 14cM 11cM which is so interesting to me.  I found one of them as  a double cousin.  Maybe some wikitree people want to work together on different DNA sites and do a beta on what works and doesn't work so we can design something new for DNA on wikitree.
by April Rarick G2G6 (8.3k points)
+6 votes
Great! But don't call it triangulation or anything that gets the technical people going. Make it a NEW relationship! I have so many cousins through Ancestry thrulines. There was an implementation of the old way that Ancestry showed common ancestors in wikitree. Call it Thruline cousins/ancestor or some such. Maybe a way to show the number of cousins through each child, the shared DNA with pairs of descendants. I tried to outline this on my profile. When you have dozens of people with shared DNA showing the same ancestor, through multiple children, to me that is strong evidence. I would like a way to show it. I have been asking for awhile.
by Sue Hall G2G6 Pilot (123k points)
I agree that having several DNA matches linking back to the same ancestor along well-documented lines can be as good an indicator of DNA confirmation as triangulation. There probably needs to be a stricter limit on the number of generations since the further back we go, the more likely pedigree collapse will muddy the waters.

https://www.wikitree.com/wiki/Shepardson-52

31 DNA matches through Daniel Shepardson. Children Jonathan, Prudence-2 lines,Millie-7 lines, Daniel-3 lines, David-14 lines, Eri-4 lines.

Please don't tell me this means nothing. By the way he was one of the DNA Circles. Why were DNA circles interesting and this is not.

 

Suggested category - DNA Ancestor

parameters: testing company, reporting wikitreer, date, number of cousins, number of children-number of lines to ancestor, min-max shared DNA for each child to wikitreer
lines through children

***Jonathan(18cm2seg),
***Prudence-2lines-(18cm1seg,21cm2seg),
***Millie-7lines,(9cm1seg,9cm,1seg8cm1seg13cm1seg,15cm1seg19cm1seg),
***Daniel-3lines(sister and 1st cousins) ,
***David-12 lines (16cm2seg,8cm1seg,8cm1seg,9cm1seg,13cm2seg,12cm1seg,12cm1seg13cm1seg,10cm1seg),
***Eri-2lines (9cm1seg,7cm1seg,12cm1seg)
+6 votes
I’ve said it before and I’ll say it again: ThruLines is unreliable at best and usually wrong at worst. Especially when multiple trees and poorly made ancestry trees come into play. Always back your stuff up with research and talking to the person. Maybe take the initiative to invite the DNA match to Wikitree. It cannot be used on its own or even with that one primary source - because so often the person is wrong somehow. I would want some definitive evidence and conclusions drawn up.
by Liz Marshall G2G6 Mach 7 (78k points)
Liz;

Thrulines in not usually wrong unless you have entered a faulty tree.

If you have entered an accurate tree, then it will identify appropriate DNA matches.

Unfortunately, there are a number of people with faulty trees and it can be very difficult to correct them.
Andrew, the problem is not with my tree, but with others. Their trees can be the faulty ones that skew the matches. Especially since Ancestry tends to take the worst sourced ones as the truth. Which is why we need to acknowledge that it is only to be used with proper contact, research, and fact checking - like ANYTHING else in genealogy.
Just because you have cousins with faulty lines does not mean that everyone does. And DNA is DNA so there is some connection. Please don't dis my lines.

Okay, but it’s a fairly universal experience once you start asking other genealogists. I’m big into the genealogy discord server (Link to G2G thread here) which has hundreds of experienced genealogists, and we frequently have discussions about how thrulines has to be proven, and taken with a grain of salt otherwise. 

I really am not impressed with what 'my other expert friends say'.  Everything can use more proof. You can deny whatever you want. Find A Grave, bad, Ancestry, bad, DNA, bad...

As an extra piece of information I think it can be pretty impressive. I think DNA trumps paper. The more that agrees the stronger the case. While problems with paper trees can be an issue and it is theoretically possible that they all point somewhere else, it is more probable that when people paper trees and DNA agree that there is something there.
Sue, If other trees on Ancestry that include your family members are accurate, then you may find you get good results from ThruLines. But I think you will agree that there are many inaccurate trees on Ancestry. Because of that, it's inappropriate to suggest that WikiTree should accept ThruLines as documentation for marking a person's relationship to his/her parents 'confirmed with DNA'. Liz's comment is correct when she says that ThruLines "cannot be used on its own". There are specific instructions for using Ancestry DNA for confirmation of parental relationships when the matching test takers are 3rd cousin or closer: https://www.wikitree.com/wiki/Help:DNA_Confirmation#AncestryDNA  I recommend following those instructions.
I am not going for confirmation only suggesting correlation.

Thrulines are as good as DNA Circles. It would be interesting to document with a category.

It is a piece of evidence that is worth noting.  Not proof, but it shouldn't be thrown out either.
+2 votes
I would like to see a different rule change in regard to DNA confirmation.  And that is to add another level of DNA confirmation, lower than DNA confirmed.

I would like to see the ability to record a DNA connection with 2 DNA matches with a radio button similar to DNA confirmation, to record the DNA matches that would make for a possible DNA confirmation when the 3rd link is found for triangulation.  

This would allow for easily adding the 3rd link for triangulation, then to change the ticker to DNA confirmed if and when the 3rd link is found for triangulation.  

The reason for this?    Allows for easier collaboration for ones working on DNA connections.
by C Fish G2G5 (5.5k points)
You can add a section to a profile describing your DNA information.  People interested in that ancestor can then see it just by reviewing the profile.

Related questions

+20 votes
18 answers
+30 votes
9 answers
+4 votes
2 answers
158 views asked Jul 8, 2019 in Genealogy Help by Rhonda Zimmerman G2G6 Pilot (143k points)
+3 votes
1 answer
+2 votes
1 answer
126 views asked Jul 10, 2018 in WikiTree Help by Christopher Culp G2G Crew (790 points)
+6 votes
1 answer
+3 votes
1 answer

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

disclaimer - terms - copyright

...