A Segment-ology TIDBIT

Bottom line up front: Triangulation and Clustering should have pretty much the same result at each company. This may allow imputing some Common Ancestors from Ancestry Clusters to the other companies and imputing some Triangulation segments from other companies to Ancestry Matches.

It may seem obvious, but it bears repeating. Your ancestry is fixed, static, unchanging. Your biological ancestors are determined at conception and cannot change. Your parents, grandparents, great grandparents, etc. remain the same no matter where you test.  Likewise, your DNA segments from Ancestors are determined at conception. These segments do not change throughout your life and are the same no matter where you test.

Therefore, the grouping methods we use should be roughly the same no matter where you test.

Your DNA segments don’t change. Segment Triangulation is based on your DNA segments. Each Triangulated Group (TG) is based on one of your DNA segments. A TG formed at any company would be the same specific segment at each of the other companies. In fact my DNA segment spreadsheet has 372 TGs formed from segments from all of the testing companies. The segments identified by each company “fit” into my TGs. There might be very slight differences among the companies, but the overall segments still fit only one way.

Your Ancestors don’t change. Take, for instance, the LEEDs method, which groups your 90-300cM Matches into four groups – one group for each of your four grandparents. No matter where you test, the four groups would be the same – one  for each of your four grandparents.

LEEDs is a special subset of Clustering. Clustering groups Matches on your Ancestors. Clusters should form on the same Ancestors, no matter which company is being used (Clustering depends on Shared Matches aka In Common With or Relatives in Common). It’s almost like a parallel universe at each company – for a given range of Match cMs, about the same Ancestor Clusters should result – based on your Ancestors. Clearly a Match who has tested at two or more companies, should show up in the same Cluster at each company. Maybe not 100 percent of the time, due to the vagaries of Shared Matching, but most of the time.

I need to try Walking The Clusters Back at each of the companies (at various cM thresholds) and see how parallel they are. I strongly suspect very strong concurrence with the larger Clusters, and large cM thresholds. Perhaps at some point, with lower cMs, the concurrence will drift away .To be continued…

Takeaway. We can Triangulate segments at 23andMe, FTDNA, and MyHeritage – giving each Match a TG for each shared segment. We can Cluster Matches at 23andme, FTDNA and MyHeritage, and note any concurrence of TG segments in these Clusters (usually one or a very few). We can determine some Common Ancestors at these companies. We can determine many more Common Ancestors at Ancestry (particularly out to 6C with ThruLines). We can Cluster at Ancestry and note any concurrence of Ancestors in these Clusters (there usually is one). Some Ancestry Matches have also tested/uploaded elsewhere, so we can determine their TGs. We can then compare Ancestry Clusters with Clusters at the other companies for congruence – allowing us to impute Common Ancestors to the Matches at other companies, and TG segments to Ancestry Matches. Maybe not in all cases, but probably in some cases.

[22BY] Segment-ology: Triangulation and Clustering Among Companies TIDBIT by Jim Bartlett 20231028