Brilliant. Thank you, Andy! You explain things so clearly. I was just at this point in a research project, where you saved me from a misunderstanding that would have wasted a lot of my time. I appreciate your clarity and insight. Great work!
Thanks for the tips. I'm still floundering a bit. I don't have any matches over 90 except for my immediate family. I found the Autocluster feature on MyHeritage and ended up with 30 clusters. I still need to dive into the report more to understand it better. I'm trying the Leeds right now with my husband's matches. One side has endogamy and it's a struggle trying to group them. At least on the other side, I can pretty much divide everyone at the great-grandparent level.
Glad I could help get you started. My wife has few matches above 90 cM. She does drop that to 40 cM and is aware of some of the challenges that come with dropping the threshold. Endogamy creates a host of problems to be sure.
I've used the Leeds Method a few times and autoclusters but I always have trouble identifying relationships within the clusters. I have one group that I know a few of the cousins within it so it's great when I get a new match that fits into that group.
Always try and start with the families that you know. Most people don't have trees on the websites so that makes determinining an "unknown" cluster very difficult.
Apologies for the question a year after this video went out, I've just rewatched it after having done a lot more clustering of my grandpa's DNA matches, I don't know if you'll even see this, but here goes anyway. I've been extracting the Shared Matches of all the matches to a spreadsheet, mainly because of the fact that on the website it doesn't give you the Shared Matches below 20cM, but obviously if you go to the matches below this value of gives you shared matches for them. Do you know any software or website that I can load this into that will give me a tree graph chart or do some clustering of the data I have, maybe even some kind of probability engine like the WATO tool, but for grouping matches and shared matches. I can't find anything, might have to build something myself if this doesn't exist, but would prefer not to. For anyone reading this is been extracting the data by cutting and pasting from the shared match page into a Google Sheets spreadsheet that converts it from the webpage into rows and columns in the spreadsheet. I've then copied this into Excel on my laptop where I've used separate tabs and formulae to mark matches that are the shared matches of one of multiple matches to save myself from having to type it all in manually (apparently there used to be a Chrome extension that did all the cutting and pasting for each match which no longer works).
The clustering can show maternal and paternal depending on who has tested from each side of your family tree. You're job will be to figure out how folks are related and then you can assign maternal / paternal to them. Have you started building a family tree on any platform?
My sisters have done their DNA on MyHeritage and Ancestry respectively , we have figured three quarters of family involving three grandparents but their are no close cousins for the other grandmother. What are we doing wrong ? Maryanne.
It's possible you're not doing anything wrong. Either the other grandmother's descendants haven't taken DNA tests, they have taken DNA tests and their results are privatized, or their are no living descendants. Which one seems most likely based on what you know from researching the descendants of that grandmother?
I know this is a year on and you may have figured out out by now, but if you haven't (and I know this isn't a foolproof way to do this, but you sound a little stuck) is that I would expand the criteria a bit, to down to 50cM or even further. I'm trying to use very distant cousins to get back to the great grandparent level and I'm going through all matches adding them to groups where they have a shared match in a group. It is a bit of an experiment but I know all the grandparents (and this is of my 100 year old grandpa) and want to try and create groups for the 8 great grandparents (which is trickier because everyone will be more distant, well most will be, more distant from each other, and everyone descended from them also has the potential to be in the group), or even bespoke groups where I have seen a group of people who being shared matches of each other but not knowing where they link into my tree, and my grandpa is from Ireland and some of these clusters of matches are from say Texas. In the end it may be that no one on the branch you don't have a group for, or even worse it could even have died off, though this is unlikely over two generations, though not impossible. One of my thoughts when this happens and I can't find any match is that it's more likely (possibly) that if the family emigrated from say Ireland it is less likely for them to have tested as I think they may anyway know their lineage and history and don't have the incentive to test as much then those who had an ancestor that emigrated to the new world so you could use that to try and make an assumption, a theory, and if you know where that is you could do some record based searching there (I appreciate you might not know this, but if you know one grandparent it might help find the original origin of the other). Hope this helps and hope you have got further. The other thing is that people continue to test so something may pop up that fills the gap.
I use clustering on Ancestry DNA to good effect. I reserved a light blue tag for Undetermined so I could tag my matches whether or not I've placed them in position in my tree. And when I have matches that I can't attribute to a branch of the family, I give it a Star and an Undetermined and add a note. On e I have a large Star cluster, I assign it a color tag, remove the Star, name it, and tag all members. My largest such cluster has over 30 members. Most have floating trees and are cross linked with their DNA match. Some clusters have 3 or more members triangulating the direction towards a possible common ancestor but most are still floaters for now. I've had some success here and there. I wish you'd talk more about the clusters and how to attribute them to a great grandparent or deeper back in time. Great video. I would note that Gedmatch is useful if you have sufficient matches in the cM range. I tried DNA Painter clustering using Gedmatch and it was a nonstarter. I'm currently exploring Excel cross matching.
@@patnoble466 You should go to RootsTech 2021 and watch anything by Diahan Southard. Her particular dotting system shows you how to do this - will save you several years of braincell twisting. I have her book, but it's written in a modular format, so the vids are easier IMHO.
The statement that the Leeds Method can be used with any platform is easier said than done. It worked well for me with Ancestry. Using the criteria of 400-90 cM to find 2nd cousins so I could determine my four grandparent lines, I had 68 people. Analysis showed only 30 of them were 2nd cousins. At 23andMe I had 36 people meeting the criteria, but only 9 were 2nd cousins. at MyHeritage 8 people met the criteria, but only 2 were 2nd cousins. FTDNA had no 2nd cousins. I suspect my numbers are not unique. Basically only Ancestry had sufficient 2nd cousins to successfully utilize the Leeds Method. However, all the of the others have better chromosome browsing tools. I conclude Leeds is great for Ancestry, but if you are using other databases you should utilize their excellent tools not Leeds.
I agree that Leeds is better suited for Ancestry than most other platforms, however, I have used it on all platforms with benefits. None of the other tools (other than clustering) divide up your matches into nice groups. Doesn't mean the other tools aren't useful, just that the Leeds has a specific purpose.
In this situation I would reduce the limit from 90cM to say 50cM, maybe even 20cM in an extreme situation, did you try this? It is annoying that Ancestry has more people who have tested with it but worse features than the other companies, but it is what it is. I'd also say that sometimes using 400cM as the upper limit can mean picking up people who are not from a distinct grandparent (it depends on the age of the person tested as to whether this is possible) so it is a bit of trial and error. If you get less than 4 distinct groups then I'd remove a group that this might be the case for and go again, hopefully some have trees that might help identify and confirm each group.
I see clusters but no idea who's side they are on. The largest cluster is on 23andme. I see 50+ matches all on X that are from China. So I am guessing there must have been a relative a long time ago that left Japan for China.
How much does Genetic Affairs cost? They want you to do a trial subscription but nowhere that I can find, least of all in Prices, does it say how much it costs.
Prices are available online. They give you a 200 credits when you register. That is enough for about three reports. It is basically introductory. For some people that may be enough. If you want to go on, a single payment of $7 gives you 700 credits. It is not expensive. If you are a MyHeritage or GEDmatch subscriber, you get it included in the subscription to their site.
It depends on how much you ask it "to do" with each request, too many variables. I think their credit system is actually nifty. Their prices are extremely reasonable considering the arm & leg everyone else wants.
Guys, thanks for all you do. I’ve got a question for ya today. My family is from Arkansas and came to here from Tennessee and Kentucky and what not... well my aunt got and Ancestry test and a Myheritage test. I think it was 2 or 3 years ago. The ancestry test said she was 3% African. Later they sent her an “updated” version that said nothing about African ancestry. The myheritage test said she was 2% Nigerian and 4% North African. We are descendants of a man named Edmond Collins who came to Arkansas from Tennessee and there is no info on him anywhere. Apparently one of his daughters looked just like an Indian. Did some research and Collins was one of the main surnames of Melungeon families of the Appalachians. Is the “black” ancestry legitament or just an error? My aunt is going to get a 3rd test from a different company and see what it says.
@@noahcartermusic2546 Yes, he's right. It's more than just b&w that intermarried. I was just re-checking my ethnic ratings and my 2% black is now totally gone in all the places! I will be glad when they keep improving on it so it stabilizes into something more trustworthy. Right now I treat ethnicity as a very basic hint and don't depend on it much because they really just haven't tested enough various type peoples yet everywhere. Too many countries are just not dna'd yet. Even France!
@@pinwheelgrl9304 I'll be interested to see where the dna business is in, say, 20 years after more people groups are tested. I don't trust percentages fully either but I wouldn't think "black" would show up if it wasn't actually there.
🔎 View 5 Ways to Filter Your DNA Matches 👉🏼 th-cam.com/video/TJBW_wtcl5o/w-d-xo.html
Brilliant. Thank you, Andy! You explain things so clearly. I was just at this point in a research project, where you saved me from a misunderstanding that would have wasted a lot of my time. I appreciate your clarity and insight. Great work!
Thanks for the tips. I'm still floundering a bit. I don't have any matches over 90 except for my immediate family. I found the Autocluster feature on MyHeritage and ended up with 30 clusters. I still need to dive into the report more to understand it better.
I'm trying the Leeds right now with my husband's matches. One side has endogamy and it's a struggle trying to group them. At least on the other side, I can pretty much divide everyone at the great-grandparent level.
Glad I could help get you started. My wife has few matches above 90 cM. She does drop that to 40 cM and is aware of some of the challenges that come with dropping the threshold.
Endogamy creates a host of problems to be sure.
I've used the Leeds Method a few times and autoclusters but I always have trouble identifying relationships within the clusters. I have one group that I know a few of the cousins within it so it's great when I get a new match that fits into that group.
Always try and start with the families that you know. Most people don't have trees on the websites so that makes determinining an "unknown" cluster very difficult.
Nice video! I like the 3D clustering, should look into that!
Please do!
@@FamilyHistoryFanatics your wish is my command, released AutoKinship yesterday! :-)
Awesome. I'll have to check this out. Too bad I didn't see this before I filmed 9 videos. I'll have to put this topic in the next video go round.
Thank you Andy!! 🧬
Any time!
Great video! Do you know of a resource to help me if I wanted to manually do a 2d cluster with my Ancestry matches?
A 2D Cluster is better known as the Leeds method. Check out this video th-cam.com/video/-74LJyjqo9c/w-d-xo.html
Apologies for the question a year after this video went out, I've just rewatched it after having done a lot more clustering of my grandpa's DNA matches, I don't know if you'll even see this, but here goes anyway. I've been extracting the Shared Matches of all the matches to a spreadsheet, mainly because of the fact that on the website it doesn't give you the Shared Matches below 20cM, but obviously if you go to the matches below this value of gives you shared matches for them. Do you know any software or website that I can load this into that will give me a tree graph chart or do some clustering of the data I have, maybe even some kind of probability engine like the WATO tool, but for grouping matches and shared matches. I can't find anything, might have to build something myself if this doesn't exist, but would prefer not to.
For anyone reading this is been extracting the data by cutting and pasting from the shared match page into a Google Sheets spreadsheet that converts it from the webpage into rows and columns in the spreadsheet. I've then copied this into Excel on my laptop where I've used separate tabs and formulae to mark matches that are the shared matches of one of multiple matches to save myself from having to type it all in manually (apparently there used to be a Chrome extension that did all the cutting and pasting for each match which no longer works).
Have you tried GDAT: www.getgmp.com/
@@FamilyHistoryFanatics Thanks the the suggestion, I'll check it out.
Does this show the paternal side of my family or only maternal,
The clustering can show maternal and paternal depending on who has tested from each side of your family tree. You're job will be to figure out how folks are related and then you can assign maternal / paternal to them. Have you started building a family tree on any platform?
What happens when your autocluster has no boxes and looks like everyone is related to everyone else?
Endogamy, you are in big trouble.
@@FamilyHistoryFanatics what to do then?
😂 I got a very similar problem ⚠️
I’ve been using 2d clustering but marking whether the matches are triangulated or not with a different colour
Good, it is very useful to know which of your matches triangulate.
My sisters have done their DNA on MyHeritage and Ancestry respectively , we have figured three quarters of family involving three grandparents but their are no close cousins for the other grandmother. What are we doing wrong ? Maryanne.
It's possible you're not doing anything wrong. Either the other grandmother's descendants haven't taken DNA tests, they have taken DNA tests and their results are privatized, or their are no living descendants. Which one seems most likely based on what you know from researching the descendants of that grandmother?
I know this is a year on and you may have figured out out by now, but if you haven't (and I know this isn't a foolproof way to do this, but you sound a little stuck) is that I would expand the criteria a bit, to down to 50cM or even further.
I'm trying to use very distant cousins to get back to the great grandparent level and I'm going through all matches adding them to groups where they have a shared match in a group. It is a bit of an experiment but I know all the grandparents (and this is of my 100 year old grandpa) and want to try and create groups for the 8 great grandparents (which is trickier because everyone will be more distant, well most will be, more distant from each other, and everyone descended from them also has the potential to be in the group), or even bespoke groups where I have seen a group of people who being shared matches of each other but not knowing where they link into my tree, and my grandpa is from Ireland and some of these clusters of matches are from say Texas.
In the end it may be that no one on the branch you don't have a group for, or even worse it could even have died off, though this is unlikely over two generations, though not impossible. One of my thoughts when this happens and I can't find any match is that it's more likely (possibly) that if the family emigrated from say Ireland it is less likely for them to have tested as I think they may anyway know their lineage and history and don't have the incentive to test as much then those who had an ancestor that emigrated to the new world so you could use that to try and make an assumption, a theory, and if you know where that is you could do some record based searching there (I appreciate you might not know this, but if you know one grandparent it might help find the original origin of the other).
Hope this helps and hope you have got further. The other thing is that people continue to test so something may pop up that fills the gap.
What if I have no matches on my maternal side?
Wait. Perhaps your maternal side had small families and few if any have tested.
Are you using clustering in your DNA research? If so, which ones do you use?
I use clustering on Ancestry DNA to good effect. I reserved a light blue tag for Undetermined so I could tag my matches whether or not I've placed them in position in my tree. And when I have matches that I can't attribute to a branch of the family, I give it a Star and an Undetermined and add a note. On e I have a large Star cluster, I assign it a color tag, remove the Star, name it, and tag all members. My largest such cluster has over 30 members. Most have floating trees and are cross linked with their DNA match. Some clusters have 3 or more members triangulating the direction towards a possible common ancestor but most are still floaters for now. I've had some success here and there. I wish you'd talk more about the clusters and how to attribute them to a great grandparent or deeper back in time. Great video. I would note that Gedmatch is useful if you have sufficient matches in the cM range. I tried DNA Painter clustering using Gedmatch and it was a nonstarter. I'm currently exploring Excel cross matching.
@@patnoble466 You should go to RootsTech 2021 and watch anything by Diahan Southard. Her particular dotting system shows you how to do this - will save you several years of braincell twisting. I have her book, but it's written in a modular format, so the vids are easier IMHO.
The statement that the Leeds Method can be used with any platform is easier said than done. It worked well for me with Ancestry. Using the criteria of 400-90 cM to find 2nd cousins so I could determine my four grandparent lines, I had 68 people. Analysis showed only 30 of them were 2nd cousins. At 23andMe I had 36 people meeting the criteria, but only 9 were 2nd cousins. at MyHeritage 8 people met the criteria, but only 2 were 2nd cousins. FTDNA had no 2nd cousins. I suspect my numbers are not unique. Basically only Ancestry had sufficient 2nd cousins to successfully utilize the Leeds Method. However, all the of the others have better chromosome browsing tools. I conclude Leeds is great for Ancestry, but if you are using other databases you should utilize their excellent tools not Leeds.
I agree that Leeds is better suited for Ancestry than most other platforms, however, I have used it on all platforms with benefits. None of the other tools (other than clustering) divide up your matches into nice groups. Doesn't mean the other tools aren't useful, just that the Leeds has a specific purpose.
In this situation I would reduce the limit from 90cM to say 50cM, maybe even 20cM in an extreme situation, did you try this? It is annoying that Ancestry has more people who have tested with it but worse features than the other companies, but it is what it is. I'd also say that sometimes using 400cM as the upper limit can mean picking up people who are not from a distinct grandparent (it depends on the age of the person tested as to whether this is possible) so it is a bit of trial and error. If you get less than 4 distinct groups then I'd remove a group that this might be the case for and go again, hopefully some have trees that might help identify and confirm each group.
I see clusters but no idea who's side they are on. The largest cluster is on 23andme. I see 50+ matches all on X that are from China. So I am guessing there must have been a relative a long time ago that left Japan for China.
If they are on the X, you can always use the X inheritance to eliminate ancestors who did not pass down an X chromosome.
How much does Genetic Affairs cost? They want you to do a trial subscription but nowhere that I can find, least of all in Prices, does it say how much it costs.
Prices are available online. They give you a 200 credits when you register. That is enough for about three reports. It is basically introductory. For some people that may be enough. If you want to go on, a single payment of $7 gives you 700 credits. It is not expensive. If you are a MyHeritage or GEDmatch subscriber, you get it included in the subscription to their site.
It depends on how much you ask it "to do" with each request, too many variables. I think their credit system is actually nifty. Their prices are extremely reasonable considering the arm & leg everyone else wants.
Thanks Alan and PinwheelGrl for responding so quickly.
Guys, thanks for all you do. I’ve got a question for ya today. My family is from Arkansas and came to here from Tennessee and Kentucky and what not... well my aunt got and Ancestry test and a Myheritage test. I think it was 2 or 3 years ago. The ancestry test said she was 3% African. Later they sent her an “updated” version that said nothing about African ancestry. The myheritage test said she was 2% Nigerian and 4% North African. We are descendants of a man named Edmond Collins who came to Arkansas from Tennessee and there is no info on him anywhere. Apparently one of his daughters looked just like an Indian. Did some research and Collins was one of the main surnames of Melungeon families of the Appalachians. Is the “black” ancestry legitament or just an error? My aunt is going to get a 3rd test from a different company and see what it says.
@420 rgb Wow, thank you for the detailed response! Thanks for taking the time. Sounds like a good topic to start studying.
@@noahcartermusic2546 Yes, he's right. It's more than just b&w that intermarried. I was just re-checking my ethnic ratings and my 2% black is now totally gone in all the places! I will be glad when they keep improving on it so it stabilizes into something more trustworthy. Right now I treat ethnicity as a very basic hint and don't depend on it much because they really just haven't tested enough various type peoples yet everywhere. Too many countries are just not dna'd yet. Even France!
@@pinwheelgrl9304 I'll be interested to see where the dna business is in, say, 20 years after more people groups are tested. I don't trust percentages fully either but I wouldn't think "black" would show up if it wasn't actually there.
Like pinwheel girl said, think of it as a basic hint. The fact that you have some other evidence pointing to the Melungeon of Appalachia adds to that.
Just got a Boucher cousin DNA match on AncestryDNA. Bouchere is French for Geisler or Geiszler.
Yay. Keep the research coming on the Geiszlers.