Let’s analyze a generic Triangulated Group (TG). There are several facets to this analysis…
Facets related to me:
1. My DNA segment – A Triangulated Group (TG) “segment” is a specific segment of my DNA. It is defined by a Chromosome number, start and end positions, and the total Mbp. The number of SNPs included and the cMs can be obtained through look-up tables on the internet (I have not done that).
2. TG Ancestor – A TG segment of my DNA first came from a specific Ancestor of mine. I’ll call this the TG Ancestor. This TG segment was passed down through descendants of the TG Ancestor, to one of my parents and then to me. All of my Ancestors who descended from the TG Ancestor also had that TG segment in their DNA. NB: the TG Ancestor started with a full Chromosome (a big segment) which he/she passed down – the TG segment was part of that larger segment/Chromosome. This original larger segment was then whittled down through the generations, but each generation had, at least, the full TG segment. NB: A segment may be passed down, one or more generations, intact (i.e. no whittling down), but the TG segment is always intact from the TG Ancestor down to you.
3. TG segment origin – The TG Ancestor received the TG segment (usually a larger segment) through a recombination process. His/her parent recombined DNA from their two parents to create a new chromosome to pass to the TG Ancestor. At this point, in our TG Ancestor, our TG segment is made up of parts from the TG Ancestor’s parent’s two chromosomes – one from each of the TG Ancestor’s grandparents. Thus, this whole TG segment did not exist, on one chromosome, in one person, before this time. The TG Ancestor is the first person to have this particular TG segment.
4. Logic – Matches who share this full TG segment, should also share this Common TG Ancestor – because this TG segment is unique to this TG Ancestor. [It can be argued through logic, that there is a possibility of this exact same segment being created in another person – granted. But the odds are extremely low, and even more distant when you consider this happening in the small subset of your DNA Matches in a TG]
Facets related to Matches and *shared* segments:
5. Cousin segments – In general, our cousins will get somewhat different segments than we do from our Common Ancestors. Apply #2 above to a Match. Our Ancestor passed down a chromosome to their children – some of it identical, some different. The DNA segments passed down through their children and their descendants to our Matches will be randomly different. What we see through a DNA test, is the overlap created by shared DNA segments – the part of our DNA from a Common Ancestor that overlaps. We might get Chr 06: 53-86Mbp and the Match may get Chr 06: 64-97Mpb – the “shared segment” is the overlap: Chr 06: 64-86Mbp. Our segments from Ancestors are rarely the same as our Matches’ segments from the same Ancestors, but
5. Cousin segments – In general, our cousins will get somewhat different segments than we do from our Common Ancestors. Apply #2 above to a Match. Our Ancestor passed down a chromosome to their children – some of it identical, some different. The DNA segments passed down through their children and their descendants to our Matches will be randomly different. What we see through a DNA test, is the overlap created by shared DNA segments – the part of our DNA from a Common Ancestor that overlaps. We might get Chr 06: 53-86Mbp and the Match may get Chr 06: 64-97Mpb – the “shared segment” is the overlap: Chr 06: 64-86Mbp. Our segments from Ancestors are rarely the same as our Matches’ segments from the same Ancestors, but the TG segment is always intact from the TG Ancestor down to your Match. Always be mindful of the difference between your own DNA segments, and a “shared segment” with a Match.
6. Cousins on the TG Ancestor – these Matches may share roughly the same amount of DNA as the full TG segment, but some may well share smaller segments. It all depends on the recombinations that occurred in the generations between the TG Ancestor and the Match. Matches in a TG are already analyzed to share at least part of the TG segment with you and with other Matches.
7. Closer Cousins – these Matches also may, or may not share the full segment. Actually close cousins may share somewhat larger segments with us – beyond the scope of the TG segment. This indicates these closer cousins share more than one TG with us and a closer Common Ancestor. However, this closer Common Ancestor would have to be a descendant of the TG Ancestor. Maybe the closer Common Ancestor would be a grandparent or Great grandparent.
8. Distant Cousins – Other Matches in the TG group may share smaller segments and be related through a more distant Ancestor of the TG Ancestor. Refer to #3 above. We could have a Match cousin related through a parent or grandparent, etc. of the TG Ancestor. In this case the Match would have only received the smaller DNA segment that was part of the full TG segment in the TG Ancestor. It is probable that Matches sharing small segments (in my case down to my 7cM threshold), could be cousins way beyond my genealogy horizon. This is particularly true with pile-up regions within a TG. The whole TG segment may come from a TG Ancestor well within my genealogy horizon, but the pile-up Matches are much more distant (or potentially false segments – a different story).
Summary – the Matches in a TG group can be cousins from many different generations, but all on the same ancestral line. The best estimated guess of relatedness to a TG segment would be a look-up of the cMs and then refer to the Shared cM Project. Generally on Chr X, the relationships may be further back.
Remember a TG segment represents your DNA – only your DNA – your DNA Matches will have a different TG.
[22BL] Segment-ology: Triangulated Group Analysis TIDBIT by Jim Bartlett 20230110
I know the number of segments typical of a 1st, 2nd, etc cousin is not established in a formal way, but seems to me, that you might have a good sense for what is “reasonable.” Can you generate a table…I’m not quite up to the math? I think it might help me sort through my matches to find the most promising ones to identify as, say, 3C, with least endogamy. (Lots of “close cousins” whose trees provide little by way of clues.) I am still plugging away at combining 6 kits to get to my 2g-grandmother, Rebecca, but got waylaid with the thought that maybe I had not given the Sideview matches enough attention.
eg. I have a KNOWN 3C1R with 223 cM and 17 segments (Ancestry). If a 3C is expected to share 75 cM (tabulated) over 3 segments (I don’t know), then this match isn’t very attractive for trying to dig out a 2nd 3C-ish relationship, right?
Kathryn, I really believe the best bet is the Shared cM Project – based on total cM. This is a huge, on-going, data collection program, with further analysis to improve the data. It includes data from thousands of people. Any guess that I might determine would be based on a sample of 1 (me). 1C are pretty easy to determine by total cM, as are 2C. Beyond that there starts to be more and more overlap in the probalities. Also I do not recomment putting too much emphasis on a precise relationship based on shared DNA (and/or the number of segments) – the data is just too variable. DNA Painter has a great tool – just enter the total cM and it lists all the possibilities and probabilities – much better than I could estimate. See https://dnapainter.com/tools/sharedcmv4 Incidentally, if you plug in 223cM, there, a relationship of 3C1R is off the chart – there is some other relationship involved to get 223cM shared DNA…. Jim
Okay, thanks, Jim. I thought you might say that. I know triangulating the segments is my best bet and I’ll keep plugging away. It’s a long game, this one.
Hi Jim, Thanks so much for an amazing blog, so informative. It has really fired my enthusiasm to use triangulated groups for various purposes so I’ve been going through all your posts and trying to learn from you. I’ve got my spreadsheet set up and started work on Chr 10 as that is an area of interest for me. I’ve already invested several days work but I do seem to be having a bit of a problem.
My parents are not related as I’ve checked on GEDMatch. I’m using MyHeritage matches primarily, but all 3 family kits were uploaded there rather than tested there. In addition, I tested my mother on Living DNA as it offered a swab test and Mum wasn’t able to spit by the time I tested her. My brother and I tested at Ancestry. I am having a real issue in creating triangulated groups as I’m finding matches who should triangulate are not doing so (false negatives?). Eg a match triangulates with a known paternal relative, so other matches who don’t triangulate with that relative are maternal. So far so good, but then I have matches that don’t triangulate with either the maternal or paternal. I’ve also found some matches with the same starting point that look like they “should” be paternal but actually match with my mother.
You’ve mentioned that the different testing chips could cause issues – could this be the case here? There is a bit of pedigree collapse further back in my mother’s line. When I cluster the matches I have an enormous mystery group on my father’s line, possibly caused by very early emigration to the US. I’m in Europe so have only 23 matches over 50cM at Ancestry and 19 matches over 50cM at MyHeritage.
I know you like to “focus on what you can do, rather than what you can’t” so I’m wondering if the answer is to ignore the segments that don’t fit just now and concentrate on getting the ones that do fit into groups? Maybe this is all part of the process, with judgement required? I’d just like some reassurance that I’m not wasting my time! Any advice would be much appreciated. Thanks. June
June, Sorry to be so long approving your post – some how I missed it. I may have tried to reply from my iPhone, and it didn’t work. A good post; and questions. Triangulation can usually be done only within each company, so there are usually no issues within each campany – they can appropriatly compare and show determine Triangulations. And once you have Triangulate Groups, and assigned them to the correct “side”, they conform to your own DNA and it’s crossover points. So TGs from one company should “fit” or “match” with TGs from another company.
Yes, set aside any “problem” segments. Often when you come back to them later, it all sorts out. If not just ignore them.
IMO, time spent Triangulating your Matches is well spent. At some point you have a map and all future Match segment fit easily into that framework. Jim
So, as an example: My segment from my grandmother on Chr 17 might run from 1 Mbp to 30 Mbp. That might be a combination of her parents, my g-grandparents. If so, then I can expect to get a subgroup representing my g-grandfather’s line and another representing my g-grandmother’s line. They are likely to cluster at opposite ends of my segment (assuming my segment has straddled a recombination point).
If I have understood your point correctly, how do you name the three TGs we might observe? There is the TG comprised of my 1Cs and there could be two additional TGs indicating (potentially) 2Cs and farther removed. I guess one might refer to these as nested TGs?
Kathryn – very good questions. And you do understand this correctly. There is a set of your segments (think TGs) for each and every generation. The trivial one is your parent’s generation – you got 46 very large segments (aka chromosomes) from them. And this relates to the first research we do with TGs – which side is each one on? The next level is the grandparent level. Let’s restrict this to the 44 autosomes (so I don’t have to tap dance around the fact that I’ve got 45 and you’ve got 46 chromosomes with tested segments). On one side, we have 22 autosomes. And the average number of crossovers per generation is 34. So… if we apply 34 crossovers to 22 chromosomes, we’ll have 56 fairly large chromosomes from our two grandparents on this side. (double these numbers to account for both sides). If we know the MRCA for each TG, we will know which grandparent they are from and can map them all to the grandparents. Two things at this level: we might have a 1C who shares all of one (or more) of these grandparent segments; depending on your Triangulation threshold (mine was 7cM), there may be several TGs in each grandparent segment. For example, I have a about 186 TGs per side (372 total) and about 56 grandparent segments – so roughly 3 or 4 TGs per grandparent segment. At the next level (Great grandparents), we’d add 34 more crossovers (to the 56 at the grandparent level) and have about 90 segments from our 4 Great grandparents (on one side) – an average of about 2 TGs per Great grandparent segment. For the next few generations we have a mix of 1 or 2 TGs per Ancestor segment; and then generally only one TG per Ancestor segment – in other words each Ancestor segment is a TG.
The choice to handle this is up to you. I personally am fixated on finding the Most Distant Common Ancestor (MDCA) for each TG. So if I see a TG that can be split, I divide it into one. I label my TGs with a letter so all the TGs on each chromosome are in alphabetical order (which also helps when I sort my 20,000-row spreadsheet different ways). For example 01S24 is on Chr 01; it starts 180Mbp (a letter for each 10Mbp, but adjustable as needed); it’s on my father’s  father’s  side. The TG just before 01S24 is 01Q24, and a few Match shared DNA segments spill over into 01S24 – it appears both TGs were in my Ancestor 74P (John CUMMINGS), but are probably split beyond that. As I determine the MDCA for each TG, it’s easy to see if/how adjacent TGs are closely related (some are, some are not). Hope this helps, Jim
Thanks Jim, as usual your comments are quite helpful in understanding somewhat complex details. While the starting and ending points of TG segments can certainly differ it seems one end or the other is very often virtually identical. This would seem to relate to descendency that may follow the same path for multiple generations by to divergence. I wonder if there is any way to estimate or group matches from this tidbit? Also the distance back to the common Ancestor for the X matches confuses me. While your suggestion that this could be further for an equal sized segment that matches on a different chromosome, I am almost sure I have seen other analysis that suggests the opposite. Unfortunately I cannot locate this.
James, Often Shared Segments tend to group near the start or end points of a TG – I see this in my spreadsheet – this reflects the *hard* start/stop in your own DNA – you cannot overlap beyond those points (except with close Matches whose DNA spans more than one TG). All or our (IBD) shared segments must fall into a real segment in our DNA – into a TG if we have identified one there.
On the X, DNA from a father is passed intact – so it is always the same for two generations – this makes those segments seem close – getting all of grandmother’s Chr X to your father to you – makes it look like one generation, when it’s really two. Please post, if you find contrary evidence. Jim