A Segment-ology TIDBIT
I’ve been explaining and discussing and arguing about Match Clusters recently. One debate concerns whether a Cluster is formed around an Ancestor (CA) or a Triangulated Group (TG). I argue that Clustering tends to result in 4 or 8 or 16 or 32 Clusters (or some other number of Ancestors in a given generation), depending on the shared segment cM threshold used. It might seem like I know the Ancestors and/or the TGs for each of my Clusters.
Confession time – I do not!
I’m working hard to determine as many as I can, but the current status is still spotty. I’m having a fairly good experience with TGs (98% of my DNA is covered by TGs); and know some CAs (over 80% of my TGs are known to the grandparent level). But I still have a long way to go on Chromosome Mapping. “Walking the Ancestors Back” on each TG is the name of that game.
I’m fairly new to Match Clustering, and as I look over that data (from my recent Cluster Matrix of AncestryDNA Matches over 20cM), I see lots of bare spots. I do see some trends, but in no way have I determined distant Ancestors (the CAs) for each of my Clusters. Nor have I determined the TGs for each of my Clusters – some Clusters have multiple TGs, and many have no TGs (after all, this Matrix is based on AncestryDNA data). It will take a while to analyze and weigh the information I’m collecting.
I’m working on a better analysis and a report of the one Cluster Matrix I’ve tried so far – stay tuned!
[22AA] Segment-ology: Confessions of a Match Clusterer TIDBIT by Jim Bartlett 20190214
Hi, Jim. I have not done a lot of segment triangulation, though I plan on doing more both because it’s a great practice and to understand cluster results better. My thought, though, is that you could have a cluster that includes people who do not share the same DNA segment. For instance, I might match a fourth cousin (4C), Abe, on chromosome 5 (for simplicity sake I’ll just state a chromosome). And, I match a 4C, Bob, on chromosome 17. Abe and Bob are descended from the same common ancestor, but they match each other on chromosome 20. (So, Abe, Bob, and I form a triangle, but we do not all 3 share the same DNA segments.) The cluster would likely include other people, too. I think it is very likely that not everyone in a cluster shares the same segment of DNA. Instead, they share the same common ancestor and are grouped together because they are matching each other.
Does this make sense? Do you agree? Is it something only further research of segment triangulation and cluster matching will prove/disprove?
I know I have a lot of “catching up” to do with your posts and understanding TG so I read a handful of articles today. Thanks for your wonderful posts!
Dana, Yes, the scenario you outline happens all the time in ICW lists. But when we combine the ICW lists of several Matches, the odds are much higher that the ICW Matches who are on most of the lists will be from the same CA. It’s not a guarantee, but it’s very helpful. The “proof” is when you find more Matches in the Cluster with the same CA. I’m having success checking Cluster Matches with small Trees and building them back & finding the CA. It’s a kick when it works.
Hi, Jim. Great point that not all clusters will even have a TG! I’ve worked on an automated cluster chart for 50 to 400 cM and have identified the common ancestor (or ancestral couple) of every cluster that has at least 3 people AND where I can place at least 2 of these cousins on my tree (which only excludes one cluster). Of course, these are just hypotheses, but they are really seeming to stand strong! I’ll try to post about it today.
Hi – welcome to my blog. You’ve really started something. I started with AncestryDNA Notes and MedBetterDNA and could see a trend among the Shared Matches. That’s when I tried Icicles, and then realized I needed to group the icicles – way too much effort, but I knew there were “pointers” in this. One pointed to the other side of a Generation 5 brick wall and I hit a gold mine, which diverted me for a while. Your system, and the subsequent Auto-Cluster tools, were the key! Thank you.
I’m not sure I want to say “not all Clusters will have a TG”. I know from my blog post on the Porcupine Chart, that not all of our ancestors will be genetic Ancestors. But if that is the case we won’t get legitimate shared segments with any Matches on that Ancestor. So I’m thinking, that if we have Clusters from Matches with shared segments with us, there should be a TG for each one. Like a Common Ancestor, the CA or TG exists, even if we cannot identify it. I agree we can have CAs without a TG, but in that case, I don’t think we’ll get a Cluster, either.
In any case, as we Walk the Clusters Back, I think this question will be answered someday.
Although I have 98% of my DNA covered with TGs (and the remainder of my DNA may be mostly little DNA slivers), I still have some with no credible CA. However, I think Clustering will provide some new leads. Jim