A Segment-ology TIDBIT
Genetic Genealogy has two main parts: genetic – the Shared DNA Segments; and genealogy – the Most Recent Common Ancestor (MRCA) with a Match. In a perfect world we link a Match and his/her Shared DNA Segment to the MRCA who passed it down to both of us.
Shared DNA Segments can be found for Matches at 23andMe, FamilyTreeDNA, MyHeritage and (by uploading our raw DNA data file) at GEDmatch. Unfortunately, none of those companies have nearly as many good Trees as Ancestry has. So finding MRCAs is hard.
Finding MRCAs is best done at AncestryDNA – many more people have tested there, and more of our Matches have good, in depth, Trees there. Unfortunately, AncestryDNA does not provide the precise Shared DNA Segment data that the other companies do.
The best outcome are Matches with MRCAs and Shared DNA Segments. I’ve run out of patience looking for MRCAs at FTDNA, MyHeritage, 23andMe and GEDmatch. Instead I am now looking for DNA segment data for my thousands of Matches at Ancestry with MRCAs.
This post will cover ways to get Segment data for AncestryDNA Matches – there are several:
1. Click on the Match name to bring up their profile – some have already uploaded to GEDmatch and list their Kit number in their profile.
2. Message the DNA Matches and request, suggest, cajole them to upload their raw DNA data to GEDmatch. I wrote a blogpost, here, about doing this. I’ve messaged many Matches requesting that they upload to GEDmatch. A few have… The best results occur when I include my email address and promise to report back my findings and to help them with autosomal DNA.
3. Ask the DNA Match if they have tested at one of the other companies, and what is their user name there. Some have… I’ve tested at all the companies and can usually find them.
4. Try to find the Ancestry user name at GEDmatch or vice versa. It sometimes works.
However, in looking at my GEDmatch One-to-Many list, I see many more Ancestry kits, that I have not yet linked to Ancestry names. Many folks use use very different names.
NB: Large segments (say over 30cM) will usually be about the same cM at AncestryDNA and the testing companies/GEDmatch. However, many segments below 30cM have been “Timbered”, and Ancestry then reports a smaller segment than the other companies report. You can always click on the “segment” line on their Match page and see what the “unweighted” cM value is – this is usually fairly comparable to what you see at GEDmatch. It’s a good idea to check this when there is an apparent discrepancy.
A better way – a Segment-ology TIDBIT
1. At GEDmatch Tier 1, run the One-to-Many list. When I set the limit to 1,000 Matches, the smallest Match shares 22.6cM – a good place to start. NB: By default, this list is sorted with the Matches with the most shared DNA at the top.
2. Sort the list on the Source column (it has the source of the DNA test data)
3. Scroll down the list to the beginning of the Ancestry kits. NB: these Ancestry Matches are still listed with the largest total cM at the top.
4. Work down this Ancestry list one by one, trying to find the Match at Ancestry. The closest ones at the top of the GEDmatch One-to-Many list are usually the easiest to find near the top of your AncestryDNA list of DNA Matches. Usually the largest Matches (most cM) will have the same total Shared DNA cM at GEDmatch and AncestryDNA – so even if the names are different, it’s often easy to find the right one at AncestryDNA.
5. As you go down the list, the AncestryDNA cM total tends to be smaller than the GEDmatch total, due to the Timber down-weighting. NB: you can always click on a Match’s AncestryDNA cM total to see what the unweighted total would be – it is usually pretty close to the GEDmatch total.
6. By working down both lists (the GEDmatch list and the Ancestry list), I’ve found they are roughly in the same order. And, through a combination of cM amount, user names and email addresses, I’ve been able to find most of the top GEDmatch Matches at Ancestry. If there is some doubt, I’ll look at the Shared Matches at Ancestry to see if any grouping would provide a clue. UPDATE: GEDmatch info puts the Match in a TG – look in that TG for other Ancestry Matches, then search Ancestry for one of those Matches and scroll down their Shared Matches for a likely link (this is generally a somewhat shorter list).
So far I’ve been able to link over 90% of my top GEDmatch kits with my Ancestry Matches. It’s easy to determine the TG at GEDmatch, and I put the TG ID in the Match Notes. Even if I cannot determine an MRCA with the Match at Ancestry, the Notes are invaluable in the Shared Match lists – they clearly form Clusters in most cases.
In just a few hours, I’ve been able to link over 100 Ancestry Matches to TGs. It will get harder as the segments get smaller and more scrolling is necessary at Ancestry to find a “fit”. But this process is worth the work, IMO, as it adds TGs to Matches at Ancestry. It adds evidence about the true ancestral line for each TG.
[22BH] Segment-ology: Segment Data for Ancestry Matches TIDBIT by Jim Bartlett 20220706
Pingback: Segment Data for Ancestry Matches 2 | segment-ology
I have been doing similar to the above for several years. This is an absolutely essential strategy, given the large size of the AncestryDNA database and the brutal lack of segment data.
I have identified almost 600 GEDmatch/AncestryDNA matches. I use one of Ancestry’s dots to tag them.
After sorting by source, looking at all the Ancestry matches, one must not forget a whole slew of older “Migration – F2 – A” matches, also from AncestryDNA.
There are several Combined matches, and I assume most of them have an Ancestry component and more committed testers, but surprisingly I have not been very successful with this combination.
Another surprise is that my second most successful group after Ancestry/GEDmatch is Ancestry and FTDNA.
Martin, Thanks for your feedback – I agree with all you points. I’m now working on Walking the Clusters Back – and over half of the Clusters include a Kit with a Triangulated Group ID (TG ID). Many Clusters have two or more TG IDs which gives a good confidence level. This lets me group many Ancestry Matches with the TGs formed from the other companies. This also provides a laser-focus on the Common Ancestor to look for among the Matches in a Cluster. Thanks again for your insights. Jim
Well,part of my problem is that I have relatively few good matches on GEDmatch and FTDNA.
Certainly nomidentifiable one’s on my mother’s side.
I have quite a lot of matches on both sides on Ancestry.
The Thrulines are nearly all good, I have plenty of work on sidelines.
Some good results on 23 and me ,(but no trees?!) and on my heritage.
These have enable me to do some good work on DNA painter.
Low rate of reply to emails is a bind
LikeLiked by 1 person
Well, my 100th GEDmatch is at 20cM, so I guess I am starting with less than 1/10 the matches.
I really wish there were a flag for people with all-colonial ancestors, and another for those who have not been able to get back to where their immigrants came from.(The county/province at least.) My TMRCAs with them are way back, because my ancestors left for a different colony.
So it would be good to exclude them from my GEDmatches.
Segments were useful to me even early on, but they have only just become of more general use across my tree as more people try something away from or addition to Ancestry. I still have to do a lot of work with clusters. And for Ancestry matches, that’s all manual now.
My matches seem to lodge their segments a lot more at MyHeritage than at GEDmatch, so I do a lot of work there.
Christopher – I use my Master spreadsheet of Matches & Shared DNA Segments in several way. With such a spreadsheet I have a master list of all DNA Matches (those with segments). I routinely add columns for special interest – sort of like dots. You could add a column and mark it for Colonial Ancestors for instance. Jim
I’ve been IDing Ancestry segments this way for a long time and for me it’s well worth it. Today I know the main segments for over 95% of my 20cM+ Ancestry matches. Two other things I do:
If I see an interesting mystery match on Ancestry I’ll ask them if they or any close relatives have tested at one of the other testing services. Or if not, if they would consider uploading to GEDmatch, as you suggest. I’ve probably asked over 2000 people on Ancestry, with a response rate of about 5%, down from a few years ago.
Also, I download my match lists from the other sites and check the more unusual surnames in the Ancestry search engine. By my estimate about 5 to 6% of 23andMe users who list surnames have also tested at Ancestry. I’m just as happy if I find the tester’s first cousin or second cousin on Ancestry.
Love the blog…
Rich, Thanks for the excellent tips. We are all learning from each other. And thanks for the positive/encouraging feedback. Jim
Have really,only managed to ID a couple this way out of many
Bob, For the ones I have trouble with, I look at the GEDmatch total cM; and then scroll down my AncestryDNA Match list to that value; then slowly continue scrolling looking for a fit. Another tip I’ve used: run the GEDmatch 1-to-Many on the troublesome kit, and see if you can determine some probable at the top of that list – then search Ancesty for those surnames. Usually, but not always, I can find the GEDmatch kit at Ancestry. Good luck, Jim
You’ve sold me… I had previously done a little of this, but it was before Ancestry had made improvements to their reporting. I only had GEDmatch IDs and emails to make the match to my AncestryDNA list. But, I still find only about 1 in 10 matches have trees of any use. I suppose, like others, I need to start building their trees for them, grrr… I wonder, does Ancestry realize they could completely dominate the DNA testing market if they provided shared segment data? Thanks for sharing your suggestions and how-tos.
Barb – Yes, a genealogist’s work is never done; and we have improvement lists for all of the companies. But that is why I publish this blog – to pass on things I’ve found to make our work a little easier. Thanks for your kind feedback. Jim
I also thought that with segments Ancestry would conquer all. And then I went out into the community talking to people about how DNA would help them. Many people who don’t really understand things wanted some privacy guarantees. Several times over the years, Ancestry has said that the absence of segments helps with this. And the Law Enforcement cases in the past few years have not helped. Strangely, it’s the people who have obeyed the law all their lives who seem most concerned.
With segments and autoclustering down to 15cM I think Ancestry really would conquer the DNA market. But then people would not need to take out subscriptions to research records. So that’s not going to happen. Not with their present pricing set up.
Thank you so much for sharing your work & research strategies! I’ve long been frustrated with Timber seemingly not recognizing about 30cM of a larger chunk that is shared by a very large group of my mother’s DNA cousins (500+ matches sharing all or a substantial part of that segment & matching each other at Family Tree DNA, for example). I’ve worked with a number of DNA cousins over the years to try & clarify some of these very murky & likely multiple connections over generations & families. I hadn’t thought to click on the Ancestry segment hyperlink & check discrepancies with ‘unweighted’ cMs. Already, I can see promising patterns to investigate further–surname group affiliations with consistently the biggest Timber chunks missing. Thanks again & best wishes for your own research! Kara
Kara, Thanks for the kind feedback. Actually I’ve found the Timber info to offer an insight – The more heavily “timbered” segment do, IMO, tend to indicate segments with more distant MRCAs. Looking at it the other way, the Triangulated Groups which seem to have the most distant MRCAs, also tend to have “timbered” segments. In several cases, it has been difficult to find closer cousins on those segments… Jim
Jim, I am not as dedicated and devoted as you, but still work at it. I love taking a good match at Ancestry with only one person in the tree and building out the tree to find the MRCA. It is easily done much of the time. I seldom get a match with trees at Ancestry; It seems most are testing for Ethnicity, and most of my new matches are younger, very young.
Caith – I, too, love the thrill of building out a Tree and finding a Common Ancestor – particulaly when that CA is also in some/many of our Shared Match Tree (or can be found by building out *their* Trees).
My new Matches seem to be all over the age map. I see many whose spouses have recently passed away, and they are trying to preserve the family Ancestry. I also see a lot of adoptees and/or folks (all ages) looking for close bio-Ancestors – another fun process to find those in their Matches’ Trees.
I’m retired, and genetic genealogy has given me a great way to give back and help others (and put some finishing touches on my own Tree). Jim
I am doing this in my epic quest to find a mysterious 2xgreatgrandmother. Two things I’ve found: If you have some really interesting DNA matches, it pays to do your own tree for the match, keeping it private on Ancestry or in your own files and for your research only. Many people don’t have public trees, or they’re very small trees or, alas, have wrong info. I am also finding that bio info, beyond birth and death dates carved directly on a gravestone, is often incorrect on FindAGrave and can lead you astray, just like relying on other people’s Ancestry trees can. I have found about 10 MRCA couples shared with several different groups of matches who all share the same chromosome segment with me this way, now filling them out to look for shared ancestors and people in the right place at the right time. A long grind.
A QUESTION: I’m starting to wonder if this 20-30 cm piece of Chromosome 13 we all share, definintely on my mom’s side, could date back to a common ancestor in Europe in the 1600s before all these people came over here. Or reflects endogamy from before they arrived. Any data on how ‘durable’ segments this size might be? I have traced many of these matches back to early 1700s, to the first immigrants in many families, but the matches do not all share the same MRCA.
Sari, Thanks for your feedback – I agree with all your points – including the “long grind”;>j
The answer to your question is YES – see this post at the ISOGG/wiki: https://isogg.org/wiki/Identical_by_descent – the graph has been discussed over and over. I think the original research was developed for a medical focus; and the x-axis is in Megabase pairs (Mbp) which are roughly equivalent to cMs. So the broad spread of data may be of help to you – it shows a good percentage from over 20 generations. So, yes, we have some percentage of Matches who probably relate to us beyond most genealogy Trees. One note of caution: at this distance, we have many Ancestors, and it’s very difficult develop a lot of evidence that points to just one (usually a famous or well known Ancestor), at the exclustion of many other. Walking the Ancestor Back on one segment is one way to do this. Jim
Yes, I also find some of the FindAGrave is incorrect. At Ancestry, I dna matched with a great nephew. The tree posted had nothing to do with him, obviously. I contacted the administrator of his tree, and she said, “Oh, that is not HIS tree. He does not even have a tree. That is MY tree. My daughter is married to him.” What! My paternal half brother posted his tree, but NOT his bio tree, but the tree of his ADOPTED father and his family with no mention of his bio father. Once, I had a close dna match with a guy and since I know my tree back through my gg grands is correct, it was obvious that the tree he posted had nothing to do with him. I contacted him and VERY tactfully told him, and he said yes, he knew because others had also contacted him to tell him something was wrong with his tree. In the past, I have had a fair number of adequate trees from my dna matches with 20 cMs; but after running the Shared Matches tool at Ancestry, I still could not see how I could possibly be related to them even after “running upstream ” some of their lines in a quick search.
Caith – yes, we have to be careful with on-line Trees. I find most are correct, but the ones that are wrong tend to draw my attention. I have to force myself to look the other way around – side-step the wrong info and focus on what I can verify. Often the verification process involves Shared Matches – a correct Tree should draw a number of grouped Matches (by Cluster, Shared Matches or segment Triangulation) to the same conclusion. Jim
Good article. Have not had the Ancestry DNA test but may if I have the time and energy to do it.
Jay, A lot depends on your objective(s). With about 22,000,000 test takers in their database – plus excellent genealogy tools and records – Ancestry is the best company for the genealogy part of genetic genealogy, IMO. Jim
What does the abbreviation TG stand for? Thanks for the helpful tips!
Sent from Yahoo Mail for iPad
Ann, TG stands for Triangulated Group. Jim
And the Triangulated Group Id is generated by doing what?
Christine – I type the TG ID in based on this blog post: https://segmentology.org/2019/02/02/standard-id-for-triangulated-groups/
Hope this helps, Jim