GEDmatch is providing relationship predictions on its site for the first time ever. Find out how you’re related to the matches you see without leaving the page.
Update Feb. 22, 2022: A newer relationship predictor, which is the first and only predictor to use data from a peer-reviewed source, is now available. I hope to update the tools at GEDmatch with the new data. The new tools now available are as follows:
- Traditional relationship predictions
- A double cousin relationship predictor (added Mar. 9)
- Relationship predictions to help validate known relatives (no population weights)
- Relationship predictions for X-DNA matches (ignoring atDNA, added Jun. 29)
GEDmatch.com has just updated its site with relationship predictions. These new relationship predictions at GEDmatch are now available on the results page of the One-to-Many tool. Currently, the feature is limited to Tier 1 users on the One-to-Many – Full Version and only on the New site, i.e. not the Classic one. This is an exciting addition to the growing suite of tools available at GEDmatch, and includes the most accurate data available for relationship prediction. Here is the methodology for generating the relative probabilities between relationships. The new relationship predictions at GEDmatch are based on this tool.
One-to-Many Tool
To see relationship predictions at GEDmatch, simply click on the total shared centiMorgans (cMs) that you share with your match. This is in the column labeled Total cM under the Autosomal section.
You can pick a value such as 2,069.9 cM, shown below.
A window will then pop up showing you all of the possible relationship types as well as the probabilities for each one. Here’s what it looks like for 2,070 cM.
Other Recently Added Tools
GEDmatch has also recently integrated tools from the fabulous site Genetic Affairs. Well known for AutoClusters, the first tool to automatically sort your DNA matches into groups, Genetic Affairs has developed many other useful tools. AutoClusters and AutoSegment are both now available at GEDmatch with a Tier 1 subscription.
If you have started using these tools already, I highly recommend them. And I do hope you enjoy the new relationship predictions at GEDmatch.
Here’s a full list of the relationship prediction tools available on this site:
If you had access to the most accurate, free, relationship predictor, would you use it? Feel free to ask a question or leave a comment. And make sure to check out these ranges of shared DNA percentages or shared centiMorgans, which are the only published values that match peer-reviewed standard deviations. Or, try a tool that lets you find the amount of an ancestor’s DNA you cover when combining multiple kits. I also have some older articles that are only on Medium.
Congrats Brit, a great accomplishment!
Brit,
This sounds wonderful because I am new to DNA analysis. Most of my families are documented since arriving in the USA. But I have a North Carolina families that requires deep dives and still archival research.
Hi, Brit. A few years back I donated my DNA to GEDMatch for genetic/genealogical studies. Some months later I got a lovely email from an Irish group. They said strands of my DNA have been found in every major Scottish Highland clan. I’m still thrilled with that info. It seems that most branches on our family tree lead back to Scotland. I’ve been there once and it felt like home. You’ll find me in western Washington state (reminds me of Scotland) playing my Celtic harp. I consider all Scots my cousins 🙂
Hi Franki. Thanks for your comment. When I was a young child my Scottish McLellan ancestry was the first one I was aware of. Once I did my tree it turned out I had far more Irish ancestors than Scottish. Some of course were Scots-Irish, but I’ve even found ancestors who moved from Cheshire to Dublin. I love Washington state. I would like to spend a lot more time there than I have so far.
Hi Franki, how did you do genetic/genealogical studies? Im trying to find my Scottish Roots. I was surprised to see how much Scottish I have. I have also been drawn to Scotland and may that is why.
Hi Franki,
I find that so interesting. I have been told my entire life that my family came to America from Ireland. In my younger years, I was obsessed with Ireland. It has become so much of an obsession that even my favorite color is green, lol. Anyhow, I have always been interested in genealogy and a few years back my husband got me a kit and a one-year sub to Ancestry. I was finally able to start investigating my family. After finding a cousin on Ancestry and with her help I traced my family back to the 1400s. It appears that on both my dad’s paternal and maternal side his family actually immigrated here from the Isle of Skye, Scotland. I have absolutely no Irish in me whatsoever.
I still feel completely and utterly drawn to anything and everything Celtic though. I have never been to Scotland or Ireland but dream of someday going. I love listening to the Celtic harp and bagpipes. I attend the Celtic festival here in town yearly. My wedding band even looks like a Celtic knot with emeralds in it. It’s amazing how the Celtic blood can sing in our veins the way it does.
Sláinte,
Shanda
Would like to see “family” cm measures or more particularly cm for unions of tested people. There is a tree-based algorithm for computing the expected cm measure of the union of match segments of descendants a particular person but if the match segments overlap there is no way I know of to compute the cm measure of that union. The union is the correct statistic since the number of tested descendants inheriting DNA can be large without implying anything about how much matching DNA the targeted ancestor is expected to have had, 1 or 100 descendants inheriting the same is about the luck of inheritance and says nothing about what the selected ancestor probably had.
I need to know how to find my grandson’s biological father. How do I do that? His mother is listed but I know her side of the family. H ow do I find his father
Hi Brenda. My first recommendation is to make sure your grandson’s DNA is tested at or uploaded to all of the major sites. I also sent you an email.
I have discovered, in the past 4-5 years that my origins were from northeast, NC free people of color. It was a surprise to me beginning circa 2003 when I found that my yDNA Haplogroup was A, M-31 and since refined with Big-Y-700. I am finding all kinds of matches with these folks and would like to refine the search to the most probable yDNA progenitor 6-10 generations ago. Who was my real 4th great grandfather? I am super unhandy at computer, being of that certain generation. I have surnames that match significantly via genetics but would like to hone in these finding further. The surname Bass appears to match closely in 16 of 23 yDNA matches and Archer appears prominently in autosomal studies. Is there a consultant who can help me?
Thanks.
Hi Dr. Cale. I would be of more help with autosomal data, although I know a 4th great-grandfather is getting pretty far back there. I have an interesting 3rd great-grandfather from whom I’m quite certain that I have no DNA matches other than close, known family. There are some good groups for Y-DNA. I would recommend the Family Tree DNA (FTDNA) forums or a FTDNA user group on social media. I hope that helps.
I sympathize with your difficult search. The lack of records on black people in those years make it very hard. I am white and have numerous black cousins on 23andME, which encourages pictures of us. I have been able to find the family groups where several black cousins have their white ancestor. Tying down the exact person is more elusive. Dr. Gates assertion that the average afro-american has 12% European ancestry seems true. That is an average of one great-grandparent…. Unfortunately much of the genealogy work for black americans seems to need to be done thru their white cousins, at least initially….
Wow! Thank you, Brit, for this astounding work! I would like to understand it better. I’ve researched a Scottish line for thirty years and regularly discover others who also are tracking the same lines; one is a lady who has for 60 years and is in poor health and her hope is to solve it before she dies. We are making some progress with the help of FTdna and GEDmatch, though none of us are very good with understanding what we are doing!
My role had been to find old men and convince them to test, often providing the y-dna test kit. I’ve found several qualified men that way, from likely common ancestors of around 200 years ago. I really don’t do much with trees but use reliable data from close cousins.
What sites should I study, what new skills must I learn? I’ve collected a small group, some who know more than I, but that group is fluid as the new data puts us in smaller and different rooms!
Also, I’m following probability as it relates to the British Enlightenment and specifically to Bayes. It has been fascinating. I found, in a very old-fashioned way, that some of my Welsh ancestors were part of that group in London. They were doing everything. Electricity, probability, life insurance, supporting the American Revolution, preaching…the main character was Richard Price, who revealed Bayes to the world. There’s much more! I hope to have a presentation on YouTube soon. It’s been a blast!
I plan to follow you. Thank you!!!
I have a mystery relative (~1867-1939) who I think is either my half great uncle (if my great-GF fathered his father (no records anywhere) or some kind of half cousin if one of the brothers of my great-GF fathered his father. I am assuming the one sister was not the parent. Is it possible to determine whether this relative’s father is my great-GF as opposed to one of the brothers? I have cM data on some of the cousins in both lines.
The father of this mystery relative had a great many children across the Caribbean. I have been in contact with a number of his descendants and no one knows anything about him. No date of birth, parentage, where he was born, etc. Thank you. This mystery is driving me crazy. Plus, the mystery relatives last living daughter (age 93) would really like to know more about her father.
I followed the instructions per the “New Feature Relationship Predictions” tool. Total cM is NOT allowing the results. Is this some kind of upgrade? If so, not interested.
Hi Bruce. Are you sure you’re in the New version, a Tier 1 member, and using the One-to-Many – Full version? It will be available in more versions eventually.
Hi Brit Nicholson. I am trying to determine how a few people are related. I looked at the above sample would you go with the highest percentage group and then figure out if its on the maternal or paternal side? I have a 3 people I am really trying to figure the connection on the tree. If you could help I would appreciate it. I have tried many things and don’t know what to do. Hopefully your tool chart will help. Thanks so much!
2nd – 3rd Cousin149 cM | 2% shared DNA Father’s side
2nd – 3rd Cousin 98 cM | 1% shared DNA Father’s side
3rd – 4th Cousin 71 cM | 1% shared DNA Father’s side
Hi Dante,
There are a few ways to try to tackle this. As genetic genealogists, we have to be willing to try every one of them. Here are some important steps to consider:
You can get closer matches, more matches who are willing to respond, or more matches with trees if you test at every site. It looks like these are AncestryDNA matches. You could also test at 23andMe and upload DNA to MyHeritage, FTDNA, GEDmatch, and LivingDNA. Even better would be to get parents/aunts/uncles to test, if possible.
Out of the three, the 149 cM match is most likely to help you. You or the person you’re helping might share a great-grandparent or 2nd great-grandparent with this match.
Clustering is a great idea. I highly recommend geneticaffairs.com. GeneticAffairs also has tools integrated with MyHeritage and GEDmatch. Using their tools, you’ll get clusters of matches. Once you’ve identified the cluster(s) of interest, you can compare the matches who have trees or who respond to you. If you find the same ancestor couple in the trees of only two of these matches, that will be a big clue.
You can send messages to as many DNA matches as possible. You won’t get a reply from many of them, but the more people you message, the more likely you are to get a reply.
I hope this helps and let me know if you have any other questions.
Hello Brit, I have down loaded the python code that you published for using the Gedmatch segments, and against MyHeritage RAW data. I have used MsExcel for years to perform missing parent phasing, full family phasing, in phase SNP from to kits, template re alignment standardisation, and use the table from Gedmatch one2one to identify common segment, all of these function rely on both kits being on exactly the same template, because the comparisons are done on a line by line basis. I have about a year ago written a set of DNA tools in Python, so when I saw your article and code offering I was very interested in how you had gone about the matching. After some head scratching I have up graded my Python version to 3.10 and imported extra modules like pandas, so I could run you code. I can’t see the segment format you list is that from the tier 1 tools? Also to use your code I will have to find the differences that Myheritage list in their RAW data files so that I can import a suitable file from say 23andme or Ancestry, FTDNA etc. As I am not extensively experience in Python I have found it easy to use a GUI graphic interface format with multiple pages, selected via buttons, and return to previous pages to navigate around and a set of buttons in each function page to progress through what I want to achieve, in the opening page this also includes find a raw data .csv to import by opening a window to find the sub director and file of interest, also when I save a df data-frame I can format it and save to a separate sub directory and file name, I was surprised not to see these types of features in your code to facilitate keeping control of events and error checking should there be a problem. I assume that your tool gives the same prediction as that now find in DNAPainter which you referred to but now has been upgraded with probability estimates for potential relationships based solely on centi Morgan cM match strength. I would like to develop a free standing tool for one2one matching like that at the free tools at Gedmatch and have made a start by using an in phase tool in excel using a 99 row look up table which also gives the nature of the line be line matching between the two sets of SNPs, or NOT flagged with an X, I use ? to signal when no calls are involved and ?? when both are results like AT-AT or CT-CT where its impossible without more kits to know who donated what, and A,C,T or G as appropriate for the in-phase result, as this is easier to do for both kits completely, I then use the one2one results from Gedmatch to identify the matching segments and isolate them. I have yet to discover how to use and calculate the cM over the 22 chromosomes in excel or Python, I notice that you use a basic xK SNP count to reduce the inclusion of small segments. I’d be interested to know the detail of Gaussian distributions, as per you previous work, and how this translates when using line by line in phase comparisons and a minimum limit on match length for segments, as two random people often match up to 60% plus just on bases where the both share AA,CC,GG DD, II or TT, and there are multiple short and very short matching segment lengths less than 5cM in related individuals.
Hi Michael,
I think the first thing we should consider is whether or not this program does what you’re looking for. I think you’re referring to the code found here. I’m pretty sure that nobody else has written code for this particular task. The reason I created this code has to do with one way that I keep track of my DNA matches. This code has saved me a lot of time. Even before DNA Painter existed, I was keeping text files organized by chromosome and then segment starting position. It’s like painting, but without the visual component. For each segment, I list my top matches, how many cM they share with me, and other important information. In a way there are two parallel lists of segments running through these text files because, of course, paternal and maternal sides both have their own copies that cover the same chromosomes and positions. So I also label each segment paternal or maternal when I can figure that out.
Is this something you’re looking to do? I don’t frequently check this code to see if anything has changed that causes it not to work. Usually when I start managing a new kit I’ll run it through the program. This saves a lot of time. But, after that, I’ll have to manually sort the paternal and maternal sides and then keep adding matches from other sites. Usually, even after much time has passed, I’m able to run it without errors. But sometimes the testing sites change a column name and I have to make adjustments. I haven’t used it in several months at this point. I made this program for myself, but figured I would offer it for free to others once I had done all of the work. I’ve had the occasional user say that it didn’t work, but something like making sure they’re using the right version of Python usually fixes it.
I’m sorry I didn’t introduce error-handling into the script. I hadn’t done a lot of Python programming when I wrote this code. And, as I was just making it for my own personal use, it would have required a lot more time and energy than I was willing to spend at the time. I’ve also had a lot of other projects that I considered to have higher priority. But I would like to make sure that it’s still working when I get the chance. If we can narrow down anything in particular that needs it, I’d be glad to make some tweaks. It’s worth noting that this program could be vastly improved upon if it incorporated the in-common-with data, which I know Evert-Jan Blom uses for some of his tools. I know that he and Jonny Perl have both done some work with segments. But what’s different about my tool is that it finds only one match per segment. The goal is to have something of a master list, which you could consult for any other DNA match and then say, “Oh, that’s the so-and-so segment. I know that they share this MRCA with me because they triangulate with so-and-so.”
To answer your question about GEDmatch, yes the input data come from the Tier 1 tool called “Segment Search.” When I start managing a new kit that I know I’ll want to use for a lot of research, I pay for a month of Tier 1 membership and I download a file from this tool for any new kits I’m interested in or for kits I haven’t run the tool on in a long time.
I hope that helps to answer at least some of your questions.
Hi,
I do not know if this feature is not working for me or If I’m doing something wrong. But the results I got is not about family but about groups like
Relationship Probabilities for 44 total shared cM
Relationship Type Probability Group
3C1R Group 14.2%
3C1R, Half-3C, Half-2C2R, 2C3R 14.2%
4C1R Group 12.7%
4C1R, Half-4C, Half-3C2R, 3C3R 12.7%
4C Group 11.8%
Is this correct. What’s the meaning of this groups?
Many thanks!
Hi Ellio,
The group probabilities are for all of the relationships combined. The line below it shows all of the relationship types in that group and it shows the individual probabilities.
It may seem as though the two probabilities are redundant, because they are in the case of 44 cM—the individual probabilities are all the same or they’re close enough that they’re assumed to be the same. But if you’re interested in how these groups work, you can look at some higher matches, like 2,000-2,300 cM. I realize that you probably don’t have any matches that high, but you can always look at another person’s match list. If you click on a kit number on your One-to-One match list, you’ll be taken to another person’s match list. You could keep doing that until you find a higher match. At that point you’ll see that the groups are broken down into categories that sometimes have very different probabilities than the others.
Also, in this relationship predictor, you can see that I split up 1st cousins and 1st cousins once removed into separate groups, while they’re usually considered to be the same group because they have the same average. But, since the probabilities are different, I wanted to show the differences.
I have a mystery relative (~1867-1939) who I think is either my half great uncle (if my great-GF is his father (no records anywhere) or possibly a half cousin if one of my great-grandfather’s brothers was the mystery relative’s father. Is it possible to determine whether this relative’s father is my great-GF as opposed to one of the brothers from cM data on some of the cousins in both my great-grandfather’s lineage and one of his brothers.
The mystery relative had a great many children across the Caribbean. I have been in contact with a number of his descendants and no one knows anything about him. No date of birth, parentage, where he was born, etc. Thank you. I am trying to help last living child(age 93) of this mystery male as she would really like to know more about her father. Time is of the essence.
Hi Marcia,
So the mystery relative might be your half-great-uncle (son of your great-grandfather). Or he might be a grandson of your 2nd great-grandparents, in which case he’d be a 1st cousin twice removed (1C2R). Or are you saying that one of your great-grandfather’s brothers is a half-brother? If the mystery relative is your half-great-uncle or a 1C2R, your shared matches would show that he’s only a descendant of your great-grandfather and not your great-grandmother. And either way, her shared matches would show that you’re a descendant of only one of her paternal grandparents.
But these are different cM groups that should show different values.
The children of this man would be either your half-1C1Rs (if he were your half-great-uncle) or your 2C1Rs (if he were your 1C2R). What’s the cM amount you share with the woman who’s his daughter? It might be such that one relationship is much more likely than the other, as found here: https://dna-sci.com/tools/brit-cim/
I take it that no other of her siblings had their DNA tested before they died? It should be possible to add up the cousins’ DNA to try to get a proxy for the mystery relative’s DNA like in these articles:
https://dna-sci.com/2020/12/01/your-siblings-have-dna-kits-too-use-them-to-their-full-potential/
https://dna-sci.com/2021/12/02/your-siblings-have-dna-kits-too-part-2-testing-the-methods/
https://dna-sci.com/2022/01/10/your-siblings-have-dna-kits-too-part-3-empirical-data/
Thanks so much, Brit. No siblings of the daughter of the mystery man had DNA done before they died. I have no siblings. None of my great-grandfather’s brothers were half-brothers that I know of.
I share 170cMs with the 93yo daughter of the mystery man (mm). My 2 first cousins (same lineage) share 296 and 164, all on A.com. I have 23andMe data from mm’s daughter and from my 3rd cousin descended from one brother of my great-grandfather. He and I are both 5th generation from the father of my great-g’father. This 3rd cousin matches to mm’s daughter at 69cMs and 2 chromosome segments, 16 and 18. I match at 170cMs with 6 ch segments- 1, 4, 9, 10, 12, 21. Can that help tell which was the father of mystery man- my great-g’father or one of his brothers? Thank you.