Gedmatch has a new Tier 1 tool called AutoKinship. This is an interesting tool that creates different trees from your DNA matches, based on AutoClusters. I put down the $10 to try this new tool at Gedmatch. Roberta Estes wrote a good article on how to use the tool. I used that when I was stuck, but preferred to try to play with it on my own.
Running, Downloading and Extracting
An important step that Roberta emphasizes is the extracting part. It is easy to run the tool at the suggested levels. It was easy for me to download the zipped file. However, opening the file doesn’t work correctly unless everything that you have downloaded is unzipped. Here is what I used:
After I extracted all the files, I opened a file that gave me AutoClusters:
This is the file minus the names. There were 19 Clusters. I recognized some of the names.
Upping the Matches to 250
I am now on my laptop and will try this again with 250 matches. This report is taking a little longer. Roberta says that there is a 10 minute limit on these reports. This time I have gone from 19 Clusters to 45. Yikes.
This time I got a more interesting analysis:
Now I have 3 AutoTrees, Common Locations and about 20 of my clusters with AutoKinship. I’ll start with a known quantity. Pink Cluster 16 has some of my well known and documented Frazer relatives:
This is near the start of the gray squares. The gray squares indicate that these clusters are inter-related. That is due to my Protestant minority ancestors in Catholic Ireland. They tended to marry others from the Church of Ireland. I have three known Frazer Lines and two known McMaster Lines in my ancestry.
The AutoKinship Tree looks like this for Cluster 16:
Things are a little mixed up, but are within the right ball park. For example, Paul is my 2nd cousin once removed, but I should be more removed than he is. Paul is actually 2nd cousin with Emily, not 1st cousin once removed as shown. Here are some of the relationships shown differently from one of my Blogs from 2018:
I’m not sure why Bill was not included. I was never sure how Shelly fit in. She has no tree at Ancestry.
I’ll use Cluster 16 as a springboard to jump to Cluster 20:
This is still in my Irish Region:
I note that the probability for this tree has an E-6 which I take to be better than the previous tree which was in the vicinity of E-12. The only person I am aware of in this AutoKinship Tree is Mel who is Emily’s daughter from the previous AutoKinship Tree. I looked for Deborah at AncestryDNA, but could not find her. I found Annette at FTDNA, but my match with her is on Chromsome 20 which on my paternal side has been out of control in that there have been so many matches on that Chromosome.
Some More Known Rathfelder Relatives on Cluster 44
Cluster 44 also has an AutoKinship Tree:
Here is the DNA tree that I have for Rathfelder:
The AutoKinship Tree is off a bit. Catherine and I should be on the same level. Anita and Inese should be a generation after me. Donna is a special case. Donna is my 2nd cousin once removed. Even my chart does not show things perfectly. That is because Donna and Iain are 1/2 second cousins with Anita and Inese. Donna and Iain are also half 1st cousin once removed with Catherine. AH shows as 4th cousin once removed to me on the AutoKinship Tree and in my DNA/Genealogy Tree.
How Does Otis Fit In?
I have Otis in this Tree:
Otis is actually in this tree twice. However, I don’t think I have a tree in it with both AH and Otis. Donna and Iain should be in this tree also under Leonhard Rathfelder. If the AutoKinship Tree is correct, then I may suspect a Schwechheimer connection for AH or a Rathfelder connection for Otis.
My Cluster 1 AutoKinship Tree
So far, I am liking the AutoKinship Trees because they give context to the AutoClusters. Also there appears to be an element of triangulation in making these Trees which gives me an additional sense that these trees are based on science. However, I after reading Roberta Estes Blog, I am not sure that the trees are based on trianulation.
This is an interesting tree. I know that my 1st cousin once removed and 2nd cousin have the correct relationships. That leaves Deb. I was able to find her on AncestryDNA. Debra and I have a shared DNA match with Rebeka:
It is difficult to figure out where the common ancestor is between Debra and myself. One pair of common ancestors is Jonathan Hatch b 1621 and Sarah Rowley. There may also be a Palmer connection in Rhode Island.
For fun, I will paint in Debra’s DNA using DNA Painter:
Debra’s DNA does not fill in any blank spots on my Chromosome 14. However, it does identify the right-hand part of Chromosome 14 in that it is Massachusetts Colonial DNA and not English Hartley DNA. I have a lot of green Hartley-Snell DNA identified, but not a lot of it is identified as to whether it is on the Hartley side or Snell side. This DNA is most assuredly on the Snell side. The top two choices for the identity of this DNA are Hatch and Palmer. Both of these connections go back to the 1600’s which is also interesting. Maury in my painted Chromosome 14 is mmg in the AutoKinship Tree and Pat is also there.
Cluster 3 AutoKinship Tree
Lee in the second row is a person that I have been following. Lee has ancestry in Colne. This is the area where my Hartleys came from. Lee also has Hartley ancestry which is not uncommon in Colne as Hartley has been one of the most popular name in that Parish over the last several centuries. The AutoKinship Tree connects myself with Lee, Geoff and Heather.
Heather tested at 23andMe. I have written to Heather and Geoff to see if they would share their Ancestry trees with me. Now, one report at the AutoKinship Utility has Lee matching Audrey:
Audrey tested at FTDNA and has a shared match with Tracey at FTDNA. Tracey at least has a partial tree with some ancestors from Brampton, Ontario:
However, it appears that this tree only covers one side of Tracey’s ancestry. However, Tracey has a shared match at FTDNA with Amy. Amy also has Parr ancestry. Here is how Amy in blue and Tracey in red match me on Chromosome 11:
Here is part of Amy’s tree:
Amy and Tracey have the common ancestors of Joseph Parr and Calra Morrison. That means that I may have shared ancestors going back from one of these two.
Creating a Tree – But it Leads to the Wrong Line
I haven’t had much luck with these trees in the past, but I will create another tree. It turns out that this connection is on a different line:
Loughead is from County Sligo where my grandmothers Frazer and Clarke families were from. That means that either my connection with Lee is through Sligo. I do notice that Lee has Clark ancestors, but as Clark is a common name, could this be a coincidence? Lee doesn’t show any ancestors in Sligo. Lee also matches my father’s 1st cousin who has no Clarke ancestry, so I tend to think that the Lee connection is on my Hartley side.
I have heard back from Heather, so there may be a lead there.
Cluster 4 AutoKinship Tree
This is also on my Hartley side, but whereas I believe the previous tree goes back to my English Hartleys, I believe that this tree goes back to my Snell side and Colonial Massachusetts. I show a close match as a 1st cousin twice removed to two people. These are actually my second cousins. That means that I should be roughly 4th cousins with the other matches. Of course, that is based on typical matching amounts. It could be that others who matched less are not showing and that these matches could go back further in time – perhaps like the 1600’s match I mentioned above.
It helps showing these trees as it gives me hope that I may find a common ancestor or ancestors. Ned has a pretty good tree at Ancestry:
I find the Shared Surname List useful. Here is one connection:
Here we are in the early 1600’s again. Ned descends from Mary Hathaway and I descend from John Hathaway. In my previous DNA/Genealogy tree, I have gone as far back as Simon Hathaway from 1711. This is at Simon’s great-grandparent level, so back another three generations.
Adding Ned to DNA Painter
This is an interesting possibility, because Sarah Cooke’s father was on the Mayflower. Here is how Ned shows up:
This is interesting because Ned shares more DNA in this area of my paternal side Chromosome 16 than my 1st cousin once removed Maury. I didn’t paint in Jed, but he would be in about the same area.
Adding Ned to My DNA/Genealogy Tree
This should be interesting. Here is my existing Hathaway DNA matching Tree:
All I have to do is go up three generations to Arthur and Sarah Cooke and then down to Ned. Here is the connection by itself:
Of course, the connection is a bit dubious, but it is what I have. I notice that there is a Mayhew in the line which may be another connection going way back.
Cluster 6 AutoKinship Tree
This tree has an E-23 probability which seems incredibly small to me. However, of some interest to me is that there re two siblings near the top of the tree and two siblings at the bottom. If correct, then I at the 3rd cousin once removed level with Kimmy and mostly 3rd cousin level with the other DNA matches.
I found Carl at Ancestry. He has a Hannah Pontus Churchill born in Virginia in 1651. I have a Hannah Churchill in my tree born 1649 apparently in Plymouth, MA as the daughter of John Churchill and Hannah Pontus. So if my information is right John Churchill born perhaps in 1620 and Hannah Pontus would be our common ancestors. However, from above, I had that my Audrey connection was on an entirely different line going back to Ireland. Time to move on to Cluster 8
Cluster 8 Tree
Here at least I have Beth, my second cousin and the probability on this tree is not as low as the previous tree. Beth and I descend from James Hartley and Annie Snell. James Hartley had one sister. Annie was from a larger family, so my guess is that this tree could be on the Snell side. I note that GPR is a new match at Gedmatch, but I was unable to make the connection to Ancestry.
Cluster 37 Tree
This is a tree that I am familiar with:
Let’s see how accurate the AutoKinship Tree is:
This shows that, at least in this case, AutoKinship had each relationship closer by one-half step. The one exception is between me and Judy. The program then guesses at different possibilities:
Tree 4 seems to get it right:
The problem here is that Robert shows as a 2C2R to me where he is a 2C1R. However, the structure of the tree is more or less right. I don’t know that any of the trees got it perfect. There were just better trees and worse ones. Although this AutoKinship Tree does not give me any new information, it gives me an idea of how the Tree works.
Summary and Conclusions
- It seems that the AutoKinship Trees help give some context to at least some of the AutoClusters
- Looking at the AutoClusters in this way gives some hope that a common ancestor could be found some some of the unidentified clusters
- Looking at the AutoClusters in terms of trees gives a fresh look at some old matches while also picking up some new matches that have been added to Gedmatch.
- The real help is also in the reaching out to those I haven’t reached out to yet to try to make genealogical connections.