A Segment-ology TIDBIT
We can group Matches several ways:
- Each Triangulated Group (TG) includes Matches who share the same Common Ancestor (CA). This is based on your DNA segment from an Ancestor, which other Matches also share. 23andMe, MyHeritage and GEDmatch all have tools for Triangulation.
- Clustering includes Matches who share multiple Shared Matches with each other – they tend to be based on the same Ancestor. The Leeds Method focuses on 4 groups representing our 4 grandparents. This is based on the probability that groups of Shared Matches will probably have the same Ancestor. When the lowest threshold is used (6cM), all of the company Matches are included and the Clusters tend to approximate a one-to-one relationship with TGs. This is a good tool to group our Matches at AncestryDNA and FamilyTreeDNA. I blogged about some Clustering programs here.
- We can also form Clusters based on ethnicity, geography, Haplogroups, etc., but, in general, these will not be as precise as TGs and Shared Match Clustering. These Clusters are, however, often very helpful in homing in on a CA.
Groups can help us in several ways:
- Everyone in a group should have the same objective: finding the CA. There is synergy in a group; and working together often results in a better outcome. One person’s Brick Wall or bio-Ancestor (vs. an NPE) may be in the Trees of other Matches in the Group.
- Close Cousins and their CAs with you provide a beacon toward the more distant CA, and limit the possibilities that would otherwise need to be explored.
- Once several Matches in a group agree on a CA, that CA line can be imputed to the other Matches. Many times I have searched a Match’s Tree for a specific Ancestor (highlighted in the Cluster), and found it! I’ve also communicated with Matches with no/small Trees and asked specifically about a surname and gotten positive/helpful responses.
- Use Clustering to form groups at FTDNA, MyHeritage and 23andMe, and use them as a basis for TGs – Triangulation goes much more quickly when you only compare segments that will probably Triangulate.
We can form Triangulated Groups at 23andMe, MyHeritage, GEDmatch, and, with a Clustering pre-start, at FamilyTreeDNA – but those companies, generally, do not offer much in the way of genealogy tools, and only a few of the Matches have robust Trees. On the other hand, AncestryDNA has a lot of good Trees, and great tools like ThruLines, but no DNA segment data – however, we can do Clustering. DNA and No Trees; OR Trees and NO DNA – it’s frustrating… So how can we merge the TGs and AncestryDNA’s Clusters?? More on this later…
BOTTOM LINE: We need both Triangulated Segments and Triangulated Genealogy to be in sync (reinforcing each other) before we can have confidence in our conclusions. One without the other is incomplete research.
[AQ] Segment-ology: Grouping Matches – Try It! TIDBIT by Jim Bartlett 20191128
segment-ology 
Jim, could you clarify the part about imputing a CA line to other matches in a TG? Until one has done the genealogical research, one doesn’t really know the generation of the CA. One of one’s matches may be descendants of one’s gg grandparents while another is a descendant of the grandparent of one of those gg grandparents. To me, the beauty of your method for grouping segments into TGs, is that one does not have to know the genealogy to do so. One can group first, and then choose which lines are best to trace. Or one can group matches into clusters, and then research one cluster at a time, expecting that the lines are likely to converge eventually.
LikeLike
Emily,
You make a good point. By “imputing a CA line” I mean the line from you back to the deepest Ancestor you are confident about. This may meander from mother to father to mother to mother, etc. – usually NOT along a surname line. The idea is to keep this “line” in your mind as you research the Match’s Tree. In my Tree I have 5 generations of HIGGINBOTHAMS on one TG. Same for my CUMMINGS line. These are easier to focus on one surname. But most TGs will be a combination of males and females. By “imputing” I don’t mean the whole line. Another analogy I use is to treat the TG as a “pointer” – and in fact that may be the better analogy, because the TG will actually “point” back further than my genealogy will allow.
Often I save time when a Match has a large Tree – I just search the Tree for a few surnames along the TG line. If that doesn’t work, I sometimes try collateral lines to my TG. As a last effort I’ll look at all the Ancestors in a Match’s Tree and focus on geography. I agree with your point that genealogy research is required (even when you get a hit, you have to vet it). Jim
LikeLike
Pingback: Walking The Clusters Back | segment-ology
Jim, you said, “AncestryDna has a lot of good trees.” For me, AncestryDna HAD a lot of good trees! Although I have ~89,000 matches, I get very few trees, or good trees. In fact, I get very few matches of late. I used to get between 50-100+ matches per day. consistently. Now, many days can go by without my getting a match, and when I do, it is about 30.
Thank you for your many contributions.
Happy Holidays.
LikeLike
Cathy, I hope you still have a lot of the good Trees at Ancestry. Two Points: 1) my statement was a comparison with the other companies; and 2) with ThruLines I’m getting Common Ancestors with Matches who have less than 10 people in the Trees – to me this is also a great Tree, because it has enough to determine a Common Ancestor. I’ve got all my segments (TGs) worked out, I now need Matches with CAs whom I might convince to upload to GEDmatch or also test elsewhere. Jim
LikeLike
So how can we merge the TGs and AncestryDNA’s Clusters? I’m looking forward to that post, Jim.
LikeLiked by 1 person
I’m working on it!
Jim Bartlett Sent from my iPhone DNA blog: http://www.segmentology.org
>
LikeLiked by 2 people