6

I've been mapping my DNA for about two years now. Part of the process is finding the common ancestors between three or more DNA donors that match a segment of shared DNA. I spend an enormous amount of time manually browsing through ancestral trees searching for a common name, when I would think that a computer could do this task for me. But I've searched for an application and haven't been able to find one that does this.

Let me explain the process in more detail. It isn't just getting an application to try and find the ancestor between person A and person B, which is a process that most family tree applications do.

First, the process starts with looking for matching surnames. But this must include looking for all variation of the surname -- so for instance, Schmidt and Smith, I've discovered can be the same surname. And there are many example of this situation that soundex doesn't seem to catch.

Secondly, once matching surnames are found, you need to look for matching given names. And if you find matching surnames that have the same given name, you need to verify this as a potential match by looking at birthdates, deathdates and place names.

But, I often come across surnames that match, but there may not be any given names -- then I have to look to see if any of the matching surnames may have resided in the same town/village/region.

Now let me bring you back to the math of this process. Most/many of my DNA matches have 256 or more end-of-line ancestors. If I have 5 matches with 256 EoL names, that's a lot of comparisons. This is not factoring in the variance of surnames and given names. In addition there is the importance of including the dates and place names as a factor that can lead to a matching common ancestor.

Thanks to lkessler's answer: the "Mirror Tree" concept is close to what I'd like to see in an application for finding common ancestors (but as I mention in the comments, the "Mirror Tree" method, using Ancestry's online service, has it's limitations and doesn't work for me). The algorithm that Ancestry uses for making this process possible is what I would hope to see in a localized application that uses one tree (or GEDCOM file).


All my data is kept in one family tree database (or GEDCOM), and I would like this "matching" software to work with a GEDCOM (either by reading it or importing it).

2

You've hit the nail on the head. The difficult thing about DNA matches is trying to figure out the common ancestor. The idea of course, is to use both genealogical research and DNA evidence to piece things together.

There are different techniques available that are better than the manual labour you are going through.

One technique is to build mirror trees. This can be done at Ancestry DNA, basically by assigning your DNA to your match and waiting for DNA hints to appear. I won't attempt to explain how to do it here, but there are many good articles on doing it. Here's a quick YouTube video on it.

But the technique I like best is to do some DNA Painting. And a great product you can use for this is Jonny Perl's DNA Painter. The goal is to paint the shared matches of your closer relatives who have tested. You must know how you're related to these people. You assign those matches to your common ancestor and mark those matches as paternal or maternal.

enter image description here

What will happen as you paint your chromosomes is that you'll start to determine who each of your segments come from. Then when you have three or more segments of a few people who match each other, you can quickly see which of your ancestors the segment belongs to. That will cut down considerably the number of names and places you'll have to look through in the other people's trees.

These two technques still are't easy, but they can help.

Alternatively, I could suggest to do it the other way around. Download all your matches into a spreadsheet. Most DNA companies include the ancestral surnames (and some ancestral places as well) of your matches. Search for all the surnames (or places) you're interested in and note which of your matches have ancestors with which surname (or place). Then look for segment matches between you and pairs of those people with common ancestral surnames (i.e. look for triangulations between you). This will greatly increase your chance of finding people whose tree you might be able to connect with.

I always go for more general methods like these that could result in multiple successes and build on themselves. Trying to find a common ancestor one segment at a time is, as you found out, the hard way.

| improve this answer | |
  • Thanks for your answer. However, there are some limitations to each of these option. First option, mirror trees (something I hadn't heard of until now) -- if I understand correctly, you need to create a new tree for each of your DNA matches... I currently have nearly 500 matches in my database with all their ancestors. I don't know for sure, but I would think Ancestry wouldn't like it if you had 500 trees. The other thing about this is, that these 500 trees are also constantly changing, and updating them each is out of the question. Based on this, option 1 won't work for me. – TJinBC May 25 '18 at 20:18
  • Option 2, DNA Painter -- a wonderful tool. However, my parents both come from an endogamous group, and it isn't possible to apply any of my cousins to either my mother's or father's side. Consequently, I don't feel that this option will work for me either. – TJinBC May 25 '18 at 20:21
  • Option3, spreadsheet -- the option that I currently use. Not only is it cumbersome and time-consuming to compare names manual, but the spreadsheet-database (the ancestors of nearly 500 people) needs constant updating because it isn't linked to my family tree (which contains all my matches and their ancestors). – TJinBC May 25 '18 at 20:28
  • I thought I'd mention an option that I used for some time, that worked pretty well. I would print out an ancestry tree of all the my matches (with details of events and place names). Then I would visually compare each of the trees. But I overlooked too many potential common ancestors because I would forget some of the names/details during the process. An example is: if I have 5 or 6 segment matches, each match has on average 256 end-of-line ancestors... do the math, that's a lot of comparisons to do manually. In fact, I would like to compare places of EoL ancestors as well, not just names – TJinBC May 25 '18 at 20:45
  • Once again, you've identified the problem. It's not an easy process. If someone could figure out a step-by-step method that works, then it could be programmed. Until then, finding the common ancestor will be as you say time consuming. – lkessler May 25 '18 at 21:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.