Where Do Clusters Fit?

BLUF*: Unknown Clusters do NOT “fit” where known Clusters fit. Try using geography or ethnicity to find a fit.

Clusters form on a Common Ancestor (usually a couple). We can form Clusters by hand by grouping shared Matches who match each other. We can use automated programs at MyHeritage, GEDmatch, Genetic Affairs, DNAGedcom Client. These auto-programs generally Cluster all Matches within a given range of cMs – say all above 60cM. Important: they will produce Clusters across your Ancestors, whether you know those Ancestors or not. Clustering does not depend on genealogy. Matches without Trees are Clustered just as easily as Matches with full Trees.

So think about the implications…  say we did an auto-run at My Heritage and got 22 Clusters. The Cluster programs cannot give us a set of Clusters from just one generation – so we would expect to be a range. In a perfect world, 16 Cluster would align with our 16 2xG grandparents which show up as 16 3xG grandparent couples. This doesn’t happen… In my experience I’ve found that 22 Clusters would shake out over a range of generations – say roughly as 2 2xG grandparent couples; 12 3xG couples; 6 4xG couples ; and 2 5xG couples. The mix gets wider/wilder as we drop the cM threshold and the number of Clusters increases. The main point here is that there will be a distribution over several generations.

Another observation is that, in general, no two Clusters will form on the same ancestral couple. Also, we find some Matches may have a relationship to more than one Cluster (as usually shown with “gray” cells in the full Cluster diagram).

So, suppose I have 22 Clusters, and “know” which ancestor couple goes with 19 of them. What about the other 3 Clusters? Maybe the Matches don’t provide enough information for me to assign the Cluster to a Common Ancestor. Maybe the Matches provide a lot of Tree information, but none matches my Tree. I have to stop and think!

I don’t know the Common Ancestor couple (CA) for these 3 Clusters. However, almost certainly, I can rule out the 19 known CAs. Circle these in my Tree! Now look at the pattern and the “bare” spots. It’s highly likely that the 3 unknown Clusters will “fit” into the bare spots. Use judgment. Also use two other clues: geography and ethnicity. See if you can tease these clues out of the Matches in each Cluster. Compare to the best knowledge of your Ancestors in the bare spots of the likely generation in your Tree.   

Summary. Unknown Clusters point to one of your Ancestor couples, just as accurately as the “known” Clusters did; it’s just that there is not enough genealogy to “make the call”.  See if other information, like geography or ethnicity, can help.

Please post in the comments section if you’ve found other ways to link Clusters to CAs.

 *Bottom Line Up Front

[22DP] Segment-ology: Where Do Clusters Fit; by Jim Bartlett 20260828

7 thoughts on “Where Do Clusters Fit?

  1. Ciao Jim. io le mie 18 corrispondenze dna solo due su mhyeritage superano soglia cluster che anno ma posso fare cluster ha mano su un foglio con altre corrispondenze dna piccole? Che già formano sempre su mhyeritage una triangolazione?

    Like

      • Si io voglio sapere se il cluster con corrispondenze che anno segmenti sopra la soglia che non entrano nel cluster mhyeritage da loro che la loro soglia e di 15cM e io nel genrneratore cluster mhyeritage ho solo due corrispondenze che superano la loro soglia per entrare nel cluster.mentre nella triangolazione non ci sono problemi e io volevo sapere se posso farl ha mano su un foglio le corrispondenze che non rientrano nel loro cluster mhyeritage. Partendo dalla corrispondenza più alta che ho di tutte le 18 corrispondenze dna cioè 23,3cM .

        Like

  2. I have a process for differentiating shared match trios which also has applications in proofreading cluster matrices — specifically by flagging or suppressing unfavorable trios. I think you might find it of interest!

    Like

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.