3.2 cM segments in triangulation (father, uncle, son and 5th cousin)

Question

Good morning,

I would like to confirm a relationship between me and a possible 6th cousin of mine.
I’ve made the following 4 tests:
- Mine (T252218 *)
- My father’s (T904124 *)
- The 5th cousin (T711935 *)
- My uncle / father’s sibling (T364304 *)

*GEDmatch kit number.

I’ve made a Chromosome Browsing (to triangulate) and checked that each one-to-one correspondent genome overlap on the following chromosomes in the following positions:
- Chr 3: from 185813936 to 187417287 - 3.0 cM
- Chr 10: from 130676566 to 131491144 - ?
- Chr 13: from 24623069 to 25904376 - 3.2 cM
- Chr 18: from 13513316 to 18208141 - 3.2 cM

These portions of DNA are shared by all four individuals (me, my father, my uncle ans our supposed 6th cousin)

Is this data enough to confirm a legitimate match between our cousin and us?

Thank you for your attention.

Kind regards,

Lopo

Comments

Why have you dropped down to such a low level?

---

Since your dad is tested, any analysis above your dad should exclude your DNA as that doesn't add any information.  I assume you have phased your kit to get the portion that comes from your mom for research on her side (assuming she is not tested too).

---

Dear Lynda,

Thank you very much for your comment. Instead of answering you I answered myself, yesterday, by mistake.
Kind regards,

Lopo

---

Thank you William. My next step will be testing my mother. I have some mysteries to solve in my mother's side, also. Kind regards.

Answer 1

Thank you for your comment. I found no matches for higher thresholds.

I found it curious that the results of TG 1 (me, my father, 6th cousin), TG 2 (me, my uncle, 6th cousin) and the TG 3 (my father, my uncle, 6th cousin) were very similar in showing matching segments in the same chromossomes and in the same sections (few differences).

As I am quite ignorant in this matter, I'm trying to find out if these results mean something, or if they mean that our possible cousin is IDS.

The goal of my investigation, is to prove that my great great grand mother is the daughter of who we think we are, whose half-sister is the 4th grandmother of this hipothetical cousin.

Kind regards,

Lopo

Comments

In order to "prove" a triangulation, the shared cM must be at minimum 7cM. Some genealogists use 10cM as their benchmark number.

Although 3 cM might give you a hint you are all related and it looks good on paper, the chance of getting false positives with anything under 7cM is a very high percentage.  No genetic genealogist I know of would use this low of a cM threshold as proof.  I have seen some use even higher numbers than the 7 and 10 I have mentioned.

What you are showing is a very strong hint, but personally I would call it inconclusive at this point.

---

Dear Emma MacBeath,
Thank you very much for your precious and professional opinion!

Me and my father are writing a book about this side of the family, and it would be very interesting to publish conclusive DNA testing data as an example of  genetic science at the service of traditional genealogy.

I guess I will have to find a closer hypothetical cousin. This source is exhausted.

Kind regards,

Lopo

---

Lopo, It can be very frustrating I know! I am closely related to our dear WikiTree leader Abby Glann and we have been unable to find a triangulation to prove it through DNA.  She is a DNA match to my brother, but on the wrong chromosome to match others we are related to.  At times I wish to pull out my hair!

My suggestion is that you do look for others in the family line with whom you can compare DNA with.  Don't give up!  Best wishes :-)

---

I won't give up, Emma.

Thank you very much for your precious help!

Kind regards :-)

---

Dear Emma,

you wrote: In order to "prove" a triangulation, the shared cM must be at minimum 7cM. Some genealogists use 10cM as their benchmark number.

That's not correct. What you and others are mixing up is percentages of IBC/IBS (and even those are based on simulations) and false positives.

Even when 99% of all small segments below a certain threshold are IBS/IBC (aka false positives), it still means that 1% is valid IBD segments (aka true positives) and thus can be used in triangulation ("proof").

Secondly it's very important to look at the number of distant relatives we do have at the various levels (2nd, 3rd, 4th, 5th etc.)

I also refer to the "Expected number of cousins" in the following link (it somehow shows up as normal text but the next link is a clickable link):

 

While there is only a 1.1% chance of finding cousin of 7th degree there are roughly 120,000 of them. That means 1320 are waiting to be discovered.

As I said before, this is just a time problem as each and everyone of us has different priorities but it's clear that Lopo is willing to invest more time into this specific case and I do believe given the above charts probabilities that he will indeed find enough other DNA kits that match in descendants of that common ancestor.

---

Hiya Andreas!  Thank you for your imput.  I am not following your logic unfortunately.  Because how can you prove that your triangulation at the lower threshold is the 1% that is valid?  I am very interested to learn.

---

Hi Emma,

let me give it a try to explain my point better. It's a two-step process. All the matches that we find at the various testing companies or at GEDmatch etc. have false positive (IBC/IBS) and true positive (IBD) segments amongst the matches.

The method to determine what is a false positive and what not is through triangulation.

The probabilities stated are just from whatever sample was used, I have seen different percentages in the case where I have done triangulation with my matches (a 45-50% probability to triangulate for matches with a threshold of 1cM and 500 SNP's).

But these probabilities are on the macro level, applied over large amounts of matches. What we are operating with is more on the micro level, with our matches which obviously have a similar ethnic background mix.

So just because the probabilities to be IDB are low doesn't mean they are zero. To me every match provides a chance to generate more insights into my own genealogy.

Like I wrote as time is precious people are optimizing by concentrating for example on matches with more than 15 cM only (I would be long finished then as I don't have many).

But every case is special. To check and identify if a group of cousins (based on paper trail) has some ancestral segments that they share it's a much higher probability to find IBD segments then when you take two random GEDmatch numbers and compare them via 1:1 (I think everyone agrees). Again, micro (proven cousins by paper documentation) vs macro (two random people on GEDmatch) approach.

Hence it makes a lot of sense for Lopo to spend time on finding matching DNA kits and see which segments triangulate. In his case the low probability of IDB by using small segments isn't a show stopper because (usually) several of them are indeed IBD for such a narrow case (based on my own experiences).

Hope this explains my point better.

---

I understand better now  what you are alluding to especially since Mags Gaulden and I had been talking about this topic this week.  But my main question is what is the process (other than a paper trail) to prove it is not a false positive.  If I understand correctly, a simple straightforward triangulation is not enough.  Or are you saying it is if you have the paper trail to back it up?

---

Triangulation is the process of proof, Emma. What is important is that you find both, ancestral segments through triangulation and your can connect them in your family tree roughly the estimated shared DNA percentage (though that varies for distant cousins due to the randomness of DNA recombination and endogamy).

The more DNA cousins can be found for the triangulated group the better (in general). But usually we don't have a link to the MRCA (most recent common ancestor) or CA (common ancestor) and due to limits in the family trees (our own or those from our DNA cousins) this turns out to be the hardest part.

The important things for triangulation is 3 independent (meaning not recently related) individuals having minimum one an overlapping (ancestral) segment.

That's proof enough that you are related.

The only challenge is to find out how you are related and that's where the family trees come into play. There are several factors that can make it complicated, endogamy is one of them (can be that your parents are related to each other or one or more of your close ancestors). Eventually everyone will find a case of endogamy, based on the principle of pedigree collapse ("Ahnenschwund").

But as we keep on adding more and more DNA cousins to our triangulated groups, including close family members to determine on which branch of our family tree the TG exists, it becomes more and more clear which region (geographically) and eventually which couple is a MRCA or even a CA. Even if we don't have a tree it makes sense to work with our DNA cousins as others might have overlapping family trees early on and we can then research together with our DNA cousins to enlarge the family tree (including side lines) and thus connect more and more of our DNA cousins to it.

That's why it's also important IMO to add matches that triangulate and are small (between us and our match). Because that doesn't mean that amongst our DNA cousins they all have small segment matches as well. Very often I found that at least two of my DNA cousins in a TG have a large match (meaning 1st or 2nd cousin) which allows us all of a sudden to advance in our TG as we can identify the MRCA that way.

Good luck everyone!

---

Thank you very much Emma and Andreas for your comments. They truly cheered me up!

I might have an opportunity to test a more distant relative of both my father and AD, who is also descendant from our common ancestor.

I am truly dedicated to proove this genetic connection between us.

You were most helpful!

I will let you know, as soon as I have more results.

PS: I've uploaded the four kits to Genesis GEDmatch, and now, according to the new thresholds, the algorithm points both me, my father, and my uncle as matches for AD.

Lopo (me): XA6425528
Diogo (father): RS5114231
Francisco (uncle): LN8142279
AD (the cousin): HN6499044

I suppose this doesn't add much to what we know.

Thank you very much!

Kind regards,

Lopo

Answer 2

Please read: https://www.wikitree.com/g2g/199470/are-you-using-audna-segments-which-are-less-than-7-cm?

Comments

Thank you very much Peter for your answer.

To get the data mentioned in my question above, I’ve used GEDmatch Chromosome Browser, with the minimum threshold of 3 cM, which I suppose it implies a minimum SNP threshold of 300.

After reading your comments in other posts, I’ve user the one-to-one comparison using 1 cM and 500 SNP as threshold between the four kits (Son (Me) :T252218; Father: T904124; Uncle: T364304; 6th Cousin: T711935) and the only common segment shared by us all is on the Chr 8 (64,530,412 - 67,561,095).

I guess this is not enough to reach any conclusion.

Thank you,

Lopo

---

Lopo, Here is more information on the subject: https://www.wikitree.com/g2g/432940/using-small-segments-in-dna-confirmation

Answer 3

Hello Lopo,

first of all thanks for providing the GEDmatch numbers, makes it easier to do my own comparisons and follow your arguments.

I've used 1 cM and 500 SNP's and came up with two segments that are shared by your father, your uncle and the 5th (6th?) cousin:

 

Chr 8, from 64-67 mBp for a minimum of 590 SNP's (630)

Chr 14, from 76-78 mBp for a minimum of 667 SNP's (670)

 

Now your father and his brother share 50% of their DNA but they also have 50% from different sides of their parents. Hence it's safe to assume that more of the 11 (for matching your father and AD) or 15 (for matching your uncle and AD) matching segments are indeed ancestral from a common ancestor.

I'd recommend to use the paid Tier 1 feature of GEDmatch which is called "Kits that match 1 or 2 kits". You can do two runs with it, once with your father and AD's kit numbers to find other kits that match with them and the second run with your uncle and AD.

That will allow you to easily find independent (meaning not close family members) people that might help you in putting one or more TG's together.

 

I do understand that you have a paper trail to this 5th/6th cousin. In such cases it should be much easier to find enough evidence given that you have family trees available to whom you can hopefully connect the other triangulated matches. Phased data is always preferred but not possible for everyone, hence we have to use what we have.

Hope this helps!

Lastly:

The use of small threshold is a subject that leads usually to two different opinions about it. Almost as good a topic for heated discussions like the ICW is equal with triangulation (which is not). References are given hence I won't further comment on this topic, my POV is been clear and stated many times over.

Comments

Getting immediately downvoted for my post makes me wonder if this is still the right place to contribute a considerable amount of time to try to help.

To whoever downvoted:

At least use the comment section to state whatever you find worth downvoting.

But I guess the anonymity of the internet is perfect for trolling around

---

I just upvoted you, because everyone who gives a positive comment with the intent of helping deserves an upvote.  Boo to whomever downvoted without leaving a comment as to why.

---

Another upvote, Andreas. I read and appreciate all of your posts.

---

Thank you both and this is why WikiTree is so unique with it's system of Thank-you's. It stimulates that members are helping others (everyone likes positive reinforcement)

---

Hello Andreas,

I also upvoted you, because of the time you took to help me (for which i am very thankful), and above all you were indeed helpful.
I will follow your suggestion and try to find common relatives throught different tests.

Pity that Tier 1 Matching Segment Search doesn't work under 5 cM.

I will work on this enthusiastically!

Kind regards,

Lopo