G2G: Wikitree standards for DNA cluster analyses

Question

Genetic data is an important and unique record that aids in proving genealogical conclusions. One method of analyzing autosomal genetic data as part of a proof summary or argument is triangulation. WikiTree has clear guidelines for using triangulation to confirm connections to distant ancestors, and there is even an automated citation maker for triangulation analysis.

But cluster analysis is another very important type of DNA analysis. It is not currently supported as a way to mark a distant connection "Confirmed with DNA," and I think that is appropriate at this time. Writing a cluster analysis can be far more complicated than describing a triangulation between the same number of kits.

However, cluster analysis is still an important type of analysis that should be included on relevant profiles. Several prominent genetic genealogists claim that they have moved entirely away from triangulation and only do clustering.

My question: does WikiTree currently have guidelines for providing cluster analyses on profiles? Like, how much information is enough to describe the analysis completely without giving away too much information about living people? Which profiles should get the full analysis, and which can have simplified analysis, and how simplified is reasonable? What sort of information is required to give a full cluster analysis (i.e., an image of an auto-clustering grid is not sufficient, since it lacks information about the amounts of shared cMs between all pairs of people, and it does not include the relevant tree information. Is a separate citation needed when providing cluster analysis on a profile, and if so, what should be in that citation? etc. etc.

I see cluster analysis mentioned on this prominent FreeSpace page:

https://www.wikitree.com/wiki/Space:Chromosome_Mapping_Examples#Cluster_analysis

but no guidance. It seems to me that if such guidance is not currently on DNA help pages, it would be very useful to work some out and include it.

Comments

FWIW, that WikiTree page on cluster analysis that you cite is not a Help page, but rather is a member-created free-space profile.

---

Thanks for noting that. I edited the post.

---

The referenced page has a lot of links, and I'm not sure where to find information on a couple of the statements I think you're making...

A) Would you please cite the places where several prominent genetic genealogists have stated that they have moved entirely away from triangulation and only do clustering?

B) I don't understand how cluster analysis could replace triangulation, if the goal is conclusive evidence of a specific ancestry, at least I don't understand how cluster analysis alone could achieve the same high level of confidence in the conclusion. Could you explain?

---

The statements I have seen were in Facebook groups, probably all in the GGTT. So I cannot give a direct link. But, for instance, I have seen Blaine Bettinger make statements to this effect. He had also noted that there are no scientific studies that triangulation provides high confidence for common ancestry between distant cousins, although that doesn’t mean that such evidence won’t be forthcoming someday. Cluster Analysis isn’t a replacement for triangulation —- just a different method.

Regardless, this post was not about triangulation, or a comparison between it and cluster analysis. It was just an observation that because cluster analysis is common, and useful, I think it would be helpful to have some Wikitree standards about it.

Answer 1

From what I've seen, cluster analysis helps focus on a small subset of matches to review their trees and conduct traditional research.  Everyone in the cluster is likely related to each other, though could be through different MRCAs and might even be related through someone not related to you.

I'm not really sure how you'd pick which specific ancestor the cluster proves.  From the help guide, it definitely seems like it'd be appropriate to change status to Confident if there was some proof from a cluster.

https://www.wikitree.com/wiki/Help:Confident

Comments

Yes, what you say, and that is one of the reasons I wrote that I think it is appropriate at this time that cluster analysis is not approved for marking “confirmed with DNA.” (Although triangulation also poses the same problem —- the MRCA may not be the one people think.)

The post was not about marking confirmed vs. confident. It was just that cluster analysis is useful, combined with other evidence, so it is and should be on many profiles already. And so it would make sense for Wikitree to have guidelines about what, exactly, is written.

---

Hello Barry, Are you willing to write a draft example of what a cluster analysis on a profile should include?

If possible, please use actual WikiTree and GEDmatch IDs.

Thanks

---

I have found genetic genealogy to be very interesting.  I am very much into the school of genetic grouping.  I have been using dnapainter.com and yourdna.family computer program for about a year and a half and the idea is to find groups of Cousins that may have various m.c.r.a. for the different matches but just one Common Ancestor for that whole group which all shares a match on a specific chromosome segment.  You can then compare which ancestor the dna came from to that segments ethnicity description and see if that gives you any additional information for genealogy research.

---

@Peter Yes, I could write something up, although it may take me a couple of weeks. Depends on how it goes.