Genetic Affairs recently introduced a new tool – AutoKinship. Evert-Jan (EJ) Blom, the developer was kind enough to step through these results with me to assure that I’m explaining things correctly. Thanks EJ!
AutoKinship automatically predicts family trees and pathways that you may be related to your matches based on how they match you and each other. Not only is this important for genealogists trying to piece our family tree together, it’s indispensable for anyone searching for unknown ancestors, beginning with parents and walking right on up the tree for the closest several generations.
Right now, the automated AutoKinship tool is limited to 23andMe profiles, but will also work as a standalone tool where users can fill in the shared DNA information for their matches. MyHeritage, 23andMe, and GEDMatch provide centiMorgan information about how your matches also match each other. Here’s a tutorial for the standalone tool.
Unfortunately, Ancestry does not provide their customers with segment information, but fortunately, you can upload a copy of your Ancestry DNA file to MyHeritage, FamilyTreeDNA or GEDmatch, for free. You’ll find step-by-step instructions, here.
Automated AutoKinship Tool
After signing into to your Genetic Affairs account, assuming you have already set up your 23andMe profile at Genetic Affairs, click on “Run AutoKinship for 23andMe.”
I manage multiple profiles at 23andMe, so I need to click on “Profiles.”
Select the correct profile if you manage multiple kits at 23andMe.
You’ll see your various options that can be run for your 23andMe kit.
If you select AutoKinship, you automatically receive an AutoCluster because AutoKinship is built on the AutoCluster functionality.
Make your selections. I recommend leaving these settings at the default, at least initially.
The default of 250 cM excludes your closest matches. You don’t want your closest matches because they will be members of too many clustered groups.
In my initial run, I made the mistake of changing the 50 cM lower threshold to 20 cM because I wanted more matches to be included. Unfortunately, the effect this had on my results was that my largest two clusters did not produce trees.
Hint: EJ states that the software tool works from the smallest cluster to the largest when producing trees. If you notice that your largest cluster, which is usually the first one displayed in the upper left hand corner (orange here), does not have associated trees, or some people are missing, that’s your clue that the AutoKinship ran out of server time to process and you need to raise either the minimum match threshold, in this case, 50 cM, or the minimum amount of DNA shared between your matches to each other, in this case, 10 cM.
You can also select between shared matches and triangulated groups. I selected shared matches, but I may well rerun this report with triangulated groups because that provides me with a great deal of even more useful information.
When you’re ready, click on the big green “you can’t miss it” Perform AutoCluster Analysis button.
Make a cup of coffee. Your report is processing. If your email doesn’t arrive, you can click on the little envelope in your Genetic Affairs profile and the report can be downloaded to your computer directly from that link.
Your Report Arrives!
You’ll receive a zip file in the email that you MUST SAVE TO YOUR COMPUTER to work correctly. You’ll see these files, but you can’t use them yet.
First, you MUST EXTRACT THE FILES from the zip file. My zip file displays the names of the file inside of the zipped file, but they are not extracted.
You must right click, as shown above, and then click on “Extract All” on a PC. Not sure what MAC users need to do but I think it autoextracts. If you click on some of the files in this article and they don’t load correctly, or say they aren’t present, that likely means:
- You either forgot to save the file in the email to your computer
- Or you failed to do the extract
The bottom two files are your normal AutoCluster visual html file and the same information in an excel file.
Click on the AutoCluster html file to activate.
Personally, I love watching the matches all fly into place in their clusters. This html file is going to be our home base, the file we’ll be operating from for all of the functions.
I have a total of 23 interrelated autoclusters. The question is, how are we all related to each other. You can read my article about AutoClusters and how they work here.
People who are members of more than one cluster are shown with those little grey squares signifying that they match people in two clusters, not just one cluster.
For example, one cluster might be my grandparents, but the second cluster might be my maternal great-great-grandfather. Membership in both clusters tells me that my matching DNA with those people in the second cluster probably descends from my great-great-grandfather. Some of the DNA matches in the first cluster assuredly also descend from that man, but some of them may descend from other related ancestors, like my maternal grandmother. It’s our job as genealogists to discern the connections, but the entire purpose of AutoKinship is to make that process much easier.
We are going to focus on the first few clusters to see what kinds of information Genetic Affairs can produce about these clusters. Notice that the first person in row 1 is related to the orange cluster, the green cluster, the purple and the brown clusters. That’s important information about that person, and also about the interrelationship of those clusters themselves and the ancestors they represent.
Remember, to be included in a grandparent cluster, that person’s DNA segment(s) must have descended from other ancestors, represented in other clusters. So you can expect one person to be found potentially in multiple clusters that serve to trace those common ancestors (and associated segments) back in time.
The AutoKinship portion of this tool creates hypothetical trees based on relationships of you to each person in the cluster, and to the other cluster members to each other.
If you’re thinking triangulation, you’re right. I selected matches, not triangulated groups which is also an option. Some people do triangulate, but some people may match each other on different segments. Right now, it’s a jumble of hints, but we’ll sort some of this out.
If you scroll down in your html file, below your cluster, and below the explanation (which you should read,) you’ll see the AutoKinship verbiage.
I want to do a quick shout-out to Brit Nicholson, the statistician that works with EJ on probabilities of relationships for this tool and describes his methodology, here.
You’ll see the AutoKinship Table that includes a link for each cluster that could be assembled into a potential tree.
Click on the cluster you wish to view.
In my case, clusters 1 through 5 are closely related to each other based on the common members in each cluster. I selected cluster 1.
Your most probable tree for that cluster will be displayed.
I’m fortunate that I recognized three of my third cousins. AutoKinship constructed a probable genetic pedigree, but I’ve overlayed what I know to be the correct pedigree.
With the exception of one person, this AutoKinship tree is accurate to the best of my knowledge. A slot for Elizabeth, the mother of William George Estes and the daughter of Joel is missing. I probably know why. I match two of my cousins with a higher than expected amount of DNA which means that I’m shown “closer” in genetic distance that I normally would be for that relationship level.
In one case, Charles and I share multiple ancestors. In the other case, I don’t know why I match Everett on so much more DNA than his brother Carl or our other cousin, Vianna. Regardless, I do.
In one other instance, there’s a half-relationship that throws a wrench into the tree. I know that, but it’s very difficult to factor half-relationships into tree building without prior knowledge.
If you continue to scroll down, you’ll see multiple options for trees for this cluster.
Below that, you’ll see a wonderful downloadable DNA matrix of how everyone in the cluster shares DNA with everyone else in the cluster.
At this point, exit from cluster one and return to your original cluster file that shows your cluster matrix.
Beneath the AutoKinship table, you’ll see AutoCluster Cluster Information.
AutoCluster Cluster Information
Click on any one of those people. I’m selecting Everett because I know how we are related.
Voila, a new cluster configuration forms.
I can see all of the people I match in common with Everett in each cluster. This tells me two things:
- Which clusters are related to this line. In particular, the orange cluster, green, red, purple, brown, magenta and dark grey clusters. If you mouse over each cell in the cluster, more information is provided.
- The little helix in each cell tells you that those two people triangulate with each other and the tester. How cool is that?!!
Note that you can display this cluster in 4 different ways.
Return again to your main autocluster page and scroll down once again.
This just might be my favorite part.
You can import chromosome segment information into DNAPainter – instructions here.
What you’ll see next is the clusters painted on your chromosomes. I love this!!!
Of course, Genetic Affairs can’t tell you which side is maternal and which is paternal. You’ll need to do that yourself after you import into DNAPainter.
Just beneath this painting, you’ll see a chart titled Chromosome segment statistics per AutoCluster cluster.
I’m only showing the first couple as an example.
Click on one of links. I’m selecting cluster 1.
Cluster 1 has painted portions of each chromosome, but I’m only displaying chromosomes 1-7 here.
Following the painting is a visual display of each overlap region by cluster, by overlapping segment on each chromosome.
You can clearly see where these segments overlap with each other!
If you select the surname enrichment option, you’ll receive two additional features in your report.
Please note that I ran this option separately at a different time, so the cluster members and clusters themselves do not necessarily correlate with the examples above.
The Enriched Surname section of your report shows surnames in common found between the matches in each specific cluster.
Keep in mind, this does NOT just mean surnames in common with YOUR surname list, assuming you’ve entered your surnames at 23andMe. (If you haven’t please do so now.) 23andMe does not support user trees, so your entered surnames are all that can be utilized when comparing information from your matches.
These are surnames that are found more than once among your matches. I’ve framed the ones in red that I recognize as being found in my tree, and I’ve framed the ones in black that I recognize as being “married in.” In other words, some people may descend through children of my ancestors who married people with that black bracketed surname.
I can tell you immediately, based on these surnames, that the first cluster is the cluster formed around my great-great-grandparents, Joel Vannoy and his wife, Phebe Crumley.
Cluster 6 is less evident, but Anderson might be connected to the Vannoy family. I’ll need to view the common matches in that cluster at 23andMe and look for additional clues.
Cluster 9 is immediately evident too. Ferverda is Hiram Ferverda, my great-grandfather and Eva Miller is his wife.
Cluster 10 is probably the Miller line as well. Indiana is a location in this case, not a surname.
Click on “Detailed Surname Table” for more information, as shown below.
Each group of people that shares any surname is shown in a table together. In this case, these three people, who I happen to know are brothers, all share these surnames. The surnames they also share with me are shown with red boxes. The other surnames are shared only with each other and no one else in the cluster. I know they aren’t shared with me because I know my tree.
While your initial reaction may be that this isn’t terribly useful, it is actually a HUGE gift. Especially if you find a cluster you aren’t familiar with.
A mystery cluster is an opportunity to break down a brick wall. This report tells you which people to view on your match list who share that surname. My first step is to use that list and see who I match in common with each person at 23andMe.
My relatives in common with my Cluster 10 matches include my close Ferverda cousins who descend from our common Miller ancestor, plus a few Miller cousins. This confirms that this cluster does indeed originate in the Miller line.
Not everyone in that cluster shares the surname Miller. That might be a good thing.
I have a long-standing brick wall with Magdalena (surname unknown) who was married to Philip Jacob Miller, my 5-times great-grandparents. My cousins through that couple, at my same generation, would be about 6th cousins.
These matches are matching me at the approximate 4th cousin level or more distantly, so it’s possible that at least some of these matches COULD be through Magdalena’s family. In that case, I certainly would not recognize the common surnames. Therefore, it’s imperative that I chase these leads. I can also adjust the matching threshold to obtain more matches, hopefully, in this cluster, and run the report again.
Are you in love with Autokinship and its associated features yet? I am!
Wow is all I can say. There’s enough in this one report to keep me busy for days, especially since 23andMe does not support a tree function in the traditional genealogical sense.
I have several matches that I have absolutely no idea how they are related to me. This helps a great deal and allows to me systematically approach tree-building or identifying ancestors.
You can see if 23andMe has predicted these relationships in the same way, but other than messaging your matches, or finding them at another vendor who does support a tree, there’s no way to know if either 23andMe’s autogenerated tree or the Genetic Affairs trees are accurate.
What Genetic Affairs provides that 23andMe does not is composite information in one place – as a group in a cluster. You don’t have to figure out who matches whom one by one and create your own matrix. (Yes, I used to do that.)
You can also import the Genetic Affairs information into DNAPainter to make further use of these segments. I’ve written about using DNAPainter, here.
Once you’ve identified how one person in any cluster connects, you’ve found your lever to unlock the identity of the ancestors whose DNA is represented in that particular cluster – and an important clue/link to associated clusters as well.
If you don’t recognize these cousins at 23andMe, look for common surnames on your DNA Relatives match list, or see if a known close relative on your maternal or paternal side matches these people found in a cluster. Click on each match at 23andMe to see if they have provided notes, surnames, locations or even a link to a tree at another vendor.
Don’t forget, you can also select the “Based on Triangulated Groups” option instead of the “Based on Shared Matches” option initially.
Run A Report
If you have tested at 23andMe, give the Genetic Affairs AutoKinship report a try.
Is it accurate for you? Have you gained insight? Identified how people are related to you? Are there any surprises?
Do you have a mystery cluster? I hope so, because an answer just might be hiding there.
If you’d like to read more about Genetic Affairs tools, click here for my free repository of Genetic Affairs articles.
I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.
Thank you so much.
DNA Purchases and Free Uploads
- FamilyTreeDNA – Y, mitochondrial and autosomal DNA testing
- MyHeritage DNA – Autosomal DNA test
- MyHeritage FREE DNA file upload – Upload your DNA file from other vendors free
- AncestryDNA – Autosomal DNA test
- 23andMe Ancestry – Autosomal DNA only, no Health
- 23andMe Ancestry Plus Health
Genealogy Products and Services
- MyHeritage FREE Tree Builder – Genealogy software for your computer
- MyHeritage Subscription with Free Trial
- Legacy Family Tree Webinars – Genealogy and DNA classes, subscription-based, some free
- Legacy Family Tree Software – Genealogy software for your computer
- Charting Companion – Charts and Reports to use with your genealogy software or FamilySearch
- RootsMagic Software – Genealogy software for your computer
- Newspapers.com – Search newspapers for your ancestors
- NewspaperArchive – Search different newspapers for your ancestors
- Genealogical.com – Lots of wonderful genealogy research books
- Legacy Tree Genealogists – Professional genealogy research