For triangulation, how large does a matching segment need to be? [closed]

+12 votes
860 views
Can I establish a triangulation based on this information -

Kathleen (A206624) shares 66 cMs with PS (A720005)

Kathleen (A206624) shares 70.4 cMs with Nancy McWard (A731925)

PS (A720005) shares 20.9 cMs with Nancy McWard (A731925)

All three match on Chromosome 13 from 98.8 to 100.5 for 3.39 cM

Is that enough for a triangulation considering the amount they share with each other?

3rd: Re-opened because I have received some good insights and hopefully a few more will comment
WikiTree profile: Kathleen Heath
closed with the note: Thoroughly discussed at this time for this stage
in WikiTree Tech by Kathleen Heath G2G6 Mach 2 (20.6k points)
closed by Kathleen Heath

Hi Kathleen,

a member of the DNA genealogy list pointed out that those 3 kit numbers aren't matching all at chromosome 13 from 98.8 to 100.5 - when I checked them I can indeed only find the following connections:

Comparing Kit A206624 (Kathleen Heath) and A720005 (*PS)

Chr Start Location End Location Centimorgans (cM) SNPs
13   93,883,418     101,659,384  12.4                         2,693

Comparing Kit A206624 (Kathleen Heath) and A731925 (Nancy McWard)

Chr Start Location End Location Centimorgans (cM) SNPs
13  98,883,134      100,559,387  3.4                           546   

 

But for:

Comparing Kit A720005 (*PS) and A731925 (Nancy McWard)

 

I can't find a connection on chromosome 13. I've used 500 SNP and 1cM, I wouldn't advise using less than 500 SNP's as it further increases the likelihood that a second difference will be accepted as a match (500 SNP means that 1 in 500 SNP's is allowed to have a different value. If we go down to 250 SNP's for eg, we're allowing 1 in 250 or 2 in 500 differences).

 

The only thing I can find is:

Chr Start Location End Location Centimorgans (cM) SNPs
13  96,309,131      97,543,805 2.0                           291    
13 99,924,117 100,567,547 1.9 260

 

Which is at 200 SNP's and 1cM. That would mean vs the previous 500 SNP's we do have indeed 2 or even 3 differences, too many to be acknowledged at the 500cM and 1cM level.

 

Sorry to say but the ancestral segment on chromosome 13 that is shown between you and the two ladies isn't showing at all between both of them.

 

That means there are two possible explanations:

 

1) Whatever was there as ancestral segments was lost. There are several big segments between one person and the other two people and none between the two people. That looks like they are lost through unfortunate recombination. But it can also be ...

2) that the paper trail is actually correct on paper but there has been a NPE (non paternal event) and as such there is no triangulation possible for that common ancestor. Unless you're saying that all lines are just the female lines only (mother-to-mother). As soon as one line goes through a father it might be one of the question marks.

 

If it's the second case then by testing more people that are on the same line of the family tree will bring ultimately clarity which of the two is the correct one.

 

Please keep us updated!

Thank you.  After everyone's input I realized that my initial search was a misunderstanding of triangulation.  I had searched GEDmatch using the tool to compare multiple kits instead of searching each tester compared to each other.  Anyway, I learned a lot from your comments and answers.

As I find others wiling to test, I'll keep you updated.
Thanks for that clarification, Kathleen. Will all are doing baby steps when it comes to this still new topic. I for sure have done my fair share of errors as well but I also learned a lot in the meantime.

Good luck for finding more DNA evidence through triangulation, I'm pretty sure you will find a lot more TG's via GEDmatch that are legimate.

The real tricky part is finding the MRCA's and CA's

3 Answers

+4 votes
 
Best answer
Hi Kathleen,

 

yours is a really interesting case, I've never seen anything like that. Based on the rules that everyone is following (minimum 1 7cM segment for a triangulation) then this isn't a proper case for a verified ancestral segment, sorry.

I've went do to 500 SNP and 5cM which is my minimum criteria I've used in the past and even though I see several segments between each of you none of them matches better than the 3.39 cM you said.

Having said that, here's my suggestion in this case (sorry, it's close to midnight here so I can't do more on it for now):

1) Follow the advice that Magnus has given and identify other kits that actually triangulate with the three of you.

2) Start with those on the chr 13 segment that is 3.39cM

3) Also check the other segments within the group of those three. It might be that they are actually below the 5cM and 500 SNP for the third person but for two of you it's a very long match in some cases (long as for me for sure, I know the Americans usually have much longer matching segments but for a European those are long).

Maybe you will find that there is another segment where all 3 of you match, but it's actually split up.

4) Check the quality of all three kits, there is an option at GEDmatch to do that. It can be that one of them maybe has too many errors in it. You can post the % of errors here for our feedback.

 

In summary I wouldn't give up on this. Like I said, the standard rules applied at all three testing companies and the guidelines that all geneticist use are saying this isn't a valid triangulation, as it's way too small. Such a small segment is only ok if there is a larger one as well, as a second segment (with the stated 7cM).

Lastly I couldn't find any info if that is indeed proven on paper. But remember that NPE's happen. If it's indeed a triangulated match proven by paper I'd assume something in the range of 6 generations of even a lot more given the small size.

Ok, very last comment. It's very likely that on one of the other matching segments between two of the three of you the second and third connection is actually lost due to a recombination effect (with one of the person who doesn't match on the same segment). The size is very small. Not sure if you have a chance to get a DNA test from an older generation, parents or aunts/uncles. That might help overcome that.

EDIT: I did a quick check for both yourself and Nancy with PS and even though I went down to 1cM and 300 SNP (not really advisable) I couldn't find any overlap between yourself and PS and only two very tiny segments (both under 2cM) for Nancy and PS.

So, for now I'd actually rule this out as a triangulation. You need to find more DNA cousins that triangulate with the three of you. It seems that if there were more ancestral segments than they are lost in all cases but one (the chromosome 13 segment) with each of you.

Goodnight!
by Andreas West G2G6 Mach 6 (60.4k points)
selected by Jeannette Saladino
I hope you slept well.  Thanks for going a step further it is very much appreciated.  I will check the test quality as I had never known that was an option or potential problem but it makes sense. Then I'll keep plugging away on Chr 13.  Thanks again.
Yes I did, thanks Kathleen. As a side effect, finding more DNA cousins for this ancestral segment (if it is indeed one) will help you guys get over the brick wall hopefully and get a much better understanding how everyone is being related to each other.

3 people is weak as we all agreed on, if you have 4 people that would be already a big improvement, especially if you have 4 that match on paper trail (I have a similar case though further back in the past).

Lastly you asked on which "close relationships" don't count in the minimum required 3 people for a triangulation. There is no strict rule but obviously the further away people are the better. I personally rule out anyone that shares 25% or more with another person. So that would be parent-child, grandparent-child, aunts, uncles, half-siblings and siblings (to name the most popular ones).

Hope this helps, please keep us informed about your further findings.
That's what I figured so I am only considering 1 person from known family group of testers.  Thanks again for everything

Andreas, I agree that "Such a small segment is only ok if there is a larger one as well, as a second segment (with the stated 7cM)." which is from what I read was this case (but it may not have been)

I believed that step 1 requires a match must exist between all 3 participants with a predicted relationship. A single match includes at least one segment with minimum 5/7 cM then all shared segments are used in determining the genetic distance. I believe FTDNA uses 1cm, and the other sites may use a minimum of 3 cM [for these additional segments].

If the smaller segments are used in predicting the relationship, then it seems makes sense to use them for triangulation, which is step 2.

In other words, I believe that if 3 people are on paper 3rd cousins 1x removed to each other and AncestryDNA predicts these people are in this range, then the 3cm segment they share more likely than not, came from 1 of 2 people. IMO, you should be able to confirm with DNA up to but not including these 2 people for each of the 3 participants.

Ken,

I've posted the matching criteria for all 3 DTC's here: http://www.wikitree.com/g2g/236781/criteria-for-matching-segments-the-big-dna-testing-companies

 

As this is a recurring topic, I hope it will be easier to find then being "buried" deep in a thread
+4 votes
About 7 cM
by Peter Roberts G2G6 Pilot (559k points)
Andreas, clarification is most helpful. I had read that Ancestry 'phased' data but couldn't get my head around how.  Math is amazing.  As I mentioned in other replies, I think my question should have been closer to "is this enough to use as a pointer when comparing other test matches" as opposed to triangulation proof.

I have other Campbell and Lyon matches which is where this group's brick wall is.  Most of the other testers that I match with have incomplete, not sourced, or no family tree.  These two are the first that I found on Ancestry and gedmatch with the same information based on sources and not other trees.  As my reworked question shows, I'm just looking for a clue, a starting point, to approach other testers with a minimum of 8 surnames to explore from that side of the tree.

Treasure your collection of parental DNA.  My brother and I did not start our family research until after both parents had passed.  I wish you great insight and clear trails into your family tree.
Third cousins once removed share on average 26.5 cM.  See

http://isogg.org/wiki/Autosomal_DNA_statistics#Table  

So 3 cM is insufficient.

Also beyond third cousins you need triangulation between distant cousins on a 7+ cM segment.
Hello Kathleen,

What are the GEDmatch IDs for the cousins you are referring to as  matches?

Thanks and sincerely, Peter

I have a third cousin once removed that I just share 4,5 cM with (if the churchbooks are true) see my DNA nonconfirmed section....

Other people has checked the church books and story told is something like 10% of persons with that relation distance don't see each other at FTDNA (below 7cM and filtered out)

from gedmatch.com

Peter,

Me (A206624) and my brother (A371475) just for reference

AncestryDNA cousin PS is gedmatch - A720005

AncestryDNA cousin Nancy McWard is gedmatch - A731925

The paper trail confirms our relationship as much as any can and more that it does for other Ancestry testers.  Thanks for any guidance.
Hello Kathleen,

Please let me know Nancy McWard's WikiTree ID.

Please add your brother's GEDmatch ID to his DNA Tests page in WikiTree.

Peter, you stated:

"Third cousins once removed share on average 26.5 cM.  See

http://isogg.org/wiki/Autosomal_DNA_statistics#Table  

So 3 cM is insufficient."

The 26.5 cM is the total average shared amount, not for a single segment. Otherwise the 16xx cM need for a parent-child would be impossible to achieve in one chromosome even ;-)

As I wrote in my answer, I see the 3cM also as insufficient but the paper trail makes the "normal rules" not fit in this case.

Peter, Nancy McWard isn't on WikiTree at this time she is just an Ancestry match.  I am still new and haven't gotten that far on my profile creations.  Thanks for the reminder to add my brother's testing information.
Andreas, I agree that using the term "phased" to explain the 5cm is probably incorrect. I believe the Algorithm you mention is the reason.  I think I picked up on someone's usage of phasing to identify IBD segments within the phasing process as the reason.
No worries, I mostly put the small correction up here for people less familiar with the term like you and me. Incorrect things spread equally fast than correct things in the current times ;-)
+3 votes
Hello Kathleen,

According to GEDmatch, you and PS share 66 cM.

If you and PS are 2nd cousins 2x removed then sharing 66 cM is sufficient to "confirm with DNA" your relationship back to and including your shared ancestral couple because according to

http://isogg.org/wiki/Autosomal_DNA_statistics

2nd cousins 2x removed (just like 3rd cousins) are expected to share on average 53.13 cM.

Sincerely, Peter
by Peter Roberts G2G6 Pilot (559k points)
Thank you for everything

Related questions

+5 votes
2 answers
105 views asked Jul 23, 2020 in Genealogy Help by Kevin Heath G2G Crew (310 points)
+6 votes
2 answers
922 views asked Aug 15, 2016 in WikiTree Tech by Amber Heath G2G Crew (530 points)
+4 votes
2 answers
176 views asked Aug 28, 2018 in Genealogy Help by Vincent Piazza G2G6 Pilot (241k points)
+10 votes
0 answers
274 views asked Oct 20, 2014 in The Tree House by Peter Roberts G2G6 Pilot (559k points)
+1 vote
2 answers
+1 vote
1 answer

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

disclaimer - terms - copyright

...