Pages may contain affiliate links.
See my disclosure page for more details.

Untangling Your DNA Tree

This is the third post in a three-part series.


Shared matches or ICW lists are meant to be an easy to use to tool for you to get more out of your DNA results.

When your family tree is "knotted," there are a number of pitfalls to using shared matches. A "knotted" tree is my phrase for anyone who shares multiple (unexpected) ancestors with their DNA matches but isn't from an endogamous population. This might mean you have cousin marriages in your (the test taker's) tree but it might happen because of cousin marriages in the matches' trees.

My entire family is southern and this situation is extremely likely if you are southern. People aren't really talking about the problems and solutions if you have a knotted tree which is the purpose of this series of posts and the webinar.

It's pretty common knowledge that working with an endogamous population can be problematic but there's actually an entire spectrum of "interrelatedness."

Having distinct branches is one end of the spectrum. Extreme endogamy is the other end of the spectrum. Your family tree can fall anywhere in-between those extremes and when you work on specific projects, they can have different degrees of interrelatedness.

The severity of the interrelatedness (how knotted your tree or a branch is) affects using DNA and certainly affects using shared matches or ICW lists.
  • You still need to be able to use your shared matches to make progress with your DNA so you need to be aware of how knotted your tree is (or be able to recognize the situation when working with a specific branch). This is the purpose of the first post. 
  • You need to know what kind of problems you are likely to have. That's the point of the second post. 
  • And of course, you need to know what to do about this (this post's purpose).

In the second post I told you the number one solution to a knotted tree, it's really the solution to using your DNA results. Let's jump in.

Dealing with problems in your DNA

Have you figured out the number one thing you need to do to deal with this knotty problem?

It is building trees.

Flesh out the test takers tree as much as possible. If you are working with a test where you are only interested in part of their tree, you still need to double-check if the other branches could potentially be interrelated. If they are, you will need to flesh out those branches as well.

When I say "flesh out" think about making lush branches. By that I mean go beyond the direct ancestors. You definitely need to research the siblings and cousins. Eventually add second cousins and you can keep going. It's not a bad idea to look for the parents of all spouses. These are where I run into the most problems with my own family.

Don't worry, you don't have to do this all at once. This is another reason to focus on a goal. Work on fleshing out the related branches (making sure to also flesh out inter-related branches you wouldn't otherwise be interested in as that is where you'll find the problems and the solutions).

You'll also need to build trees for matches and flesh them out as appropriate (once again, you can focus only on the branches of interest as long as you've figured out the other branches are unrelated).

With the trees of matches, you can stop when you find a shared ancestor as long as you aren't potentially leaving other related branches unexplored.

A word about reality.

I know we are all limited on time. In most cases you won't have time to fully research a tree or you'll get stuck and can't get farther.

You have to make choices about what to do first and what to intentionally put off until later or until it's the only option left.

That's why we take notes.

Here's how the decision making works. Note this is very different than the decision making process for building your own tree with traditional research.

Your goal is to not go down the wrong path using shared matches if the test taker is related to a match in more than one way. This is very different than a traditional goal when researching your tree.

With DNA, in this scenario, the goal is to find all the shared ancestors, regardless if they are the source of the shared DNA. Determining which ancestor is the source of the DNA is the eventual goal. It is much harder than identifying all the ancestors shared between your trees. If you have an unidentified shared ancestor, that can complicate identifying the source of the shared DNA.

Be reasonable with this, though. When I say "all the shared ancestors," you might have a lot over 10 generations back. You can have pileups where there's a shred ancestor much farther back than you'd expect. Otherwise, you use the amount of shared DNA to consider a reasonable number of generations. You are given a lot of information with DNA results and you need to learn to use it.

Back to the decision making process. Sometimes making decisions to identify potential shared ancestors is easy.

For example, I'm all southern (all from Georgia, even). I don't work on branches for recent immigrants or from the northeast. I know which locations southerners are most likely to have moved to and I start with those (I wouldn't start with a Wisconsin line if they had a California line but I'd start with the Texas line if I could).

Based on the tree for the test-taker I'm working with, I can make decisions.

Remember, all of this is about chance. It is all based on statistics. Yes, it is possible you've made a best-guess and missed the branch where the DNA is actually shared.

Combat this by using many matches.

You wouldn't spend three years building a tree for one match. Build "quick and dirty" trees. Once you identify a very likely branch or shared ancestor, you can improve that targetted research. To start with, you want trees for as many matches as possible.

Do what you can, make estimates, not guesstimates. Make note of what you decided to do and what potential issues you couldn't deal with in a reasonable amount of time. Then move on to the next match, it might be more helpful.

Taking your DNA further

Your primary tool for working with DNA should be building trees. It doesn't matter if you have nice distinct branches or horribly knotted ones.

The more there is confusion-causing intermarriage, the more necessary it is for you to build trees to see what the situation is.

After that, you determine your next steps based on your specific goal and the situation as well as where the matches are (i.e. what company/site).

I recommend everyone do some basic steps (which you may have already done). I describe these in my free course, "The Road to DNA Success."

Here are some general steps I find most helpful when dealing with southern DNA. They should apply to any situation "in-between" normal and endogamous populations.

Untangling the Branches of Your DNA Tree

I've written before about my preferred method of clustering which is my 4 Buckets Technique. Auto-clustering tools use shared matches and shared matches are where people with too much inter-relatedness get into trouble.

Note: you can use auto-cluster tools but they are several steps down my list. My method is designed to overcome the shared matches problems but it's not something the average test taker can do (it's done in Excel and requires more Excel skills than the average test taker has---plus I'm not sure I've mastered explaining how to use the results so even if you have the Excel skills, you still have nothing without analysis).

I personally bucket southern results first before I use other tools (that's after building trees and doing what's described in "The Road to DNA Success" but before other "tools").

My goal with bucketing applies as a general goal, though.

The point of bucketing and why it's the FOUR BUCKETS technique is I wanted to generally sort the matches into the most likely correct branch. I knew I might have some errors so I didn't go for identifying exact shared ancestors or even narrowing the branch too much. If you can separate the matches into the grandparent branch (i.e. the four main branches of a family tree), that's a place to start.

If you descend from one endogamous population you might be struggling to separate into maternal and paternal. I don't know at what level of severity non-endogamous populations would have to forgo four buckets for two but the idea is to aim for knowing which grandparent branch all your closer matches belong (I can usually get my great-aunt's matches through the closer 4th cousin matches bucketed. I've had client projects with other types of "knotted" trees where it was far more random how far I could bucket matches based on the match's tree).

For a tree with distinct branches, separating into the four grandparent buckets is a pretty easy task. It can be baffling to do by hand (i.e by looking at shared matches, only) if you have confusing inter-marriages (this is exactly when inter-marriages in the matches' trees can be your problem).

What to do next depends on which site you are using.

I have a whole process I use at AncestryDNA since segment data is not available (but I have very good luck because there are so many more trees---some of which I had to add on to, of course). I'll be offering a course teaching my exact process this spring. The prerequisite is the (free) "The Road to DNA Success" so if you are interested, start there. The worst that will happen is you'll figure it out on your own while you wait!

At MyHeritageDNA I use the triangulation information and DNAPainter to try and trump the shared matches problem as much as possible.

At FamilyTreeDNA I tend to use ADSA (from DNA GEDCOM) to help me focus. There isn't a usual next step from this. I might add matches to DNAPainter or I might look for trees or build trees or look for matches also at another site.

I don't have one next step at GEDmatch because it depends on what I've found from the other sites which gives me some type of goal, even if it's a short-term goal.
  • I might be at GEDmatch looking for specific AncestryDNA matches who have transferred and then I might look at shared segments or triangulation. 
  • I might use ADSA so I can compare (quickly) to FTDNA matches. 
  • I might want to paint some chromosomes based on findings at MyHeritageDNA or use shared segments/triangulation like I would with AncestryDNA results (since MyHeritageDNA also has good trees). 

These are just the common things I do at GEDmatch. I might also use another tool based on my goal.

Just recently I have been using Double Match Triangulator (DMT). I find it works much better than auto-clustering for knotted trees. The reason is, it uses triangulation, not just shared matches. Remember, triangulation is shared matches who also share a segment of DNA.

That means DMT will miss some shared ancestors but it won't lead you down the wrong path due to a coincidental shared match group.

Unlike most auto-cluster tools, DMT allows you to enter the known relationships including a known branch but not a known shared ancestor. That is a huge deal if you have confusion between branches but it's limited to adjacent branches. It also allows you to use your discoveries that aren't confirmed new ancestors.

(For example, I have a "bucket" for a group of matches that are interrelated. I have identified who is likely a shared ancestor but I haven't confirmed his relationship. There are also some other matches in the bucket that don't descend from this person but I'm seeing the same surnames and locations in the early 1800s.

The other matches might be for his wife, or his ancestors or this is another case of cousins migrating and inter-marrying later. I know from my extensive analysis all of this is on my great-aunt's maternal side. I can estimate the branch for the matches descending from the new ancestor and still enter something for the other matches in that bucket.)

DMT is not the first tool you want to use. Build those trees. None of the tools will help without building trees. The more estimates of relationships or branches you put in DMT, the more useful the output.

The big draw with it for me is, it helps separate out the people who share DNA versus those who just share multiple ancestors (i.e. it overcomes the pitfall of false shared match groups and can indicate which shared ancestor the shared DNA is from). I have a huge number of quick and dirty trees to use to do the analysis. Without having built trees, I'd just have another Excel spreadsheet.

I would still do my Four Bucket Technique before using DMT but DMT is easy to use like an auto-cluster tool. It still can't deal with the complexity of being related to matches in multiple ways but I find it more useful than the auto-cluster tools. If you have distinct branches or little knotting in your tree, auto-cluster tools should work better as they won't miss those "bonus" shared match group clues.

Remember, for a non-knotted tree, shared matches provide more leads. With a knotted tree, those extra leads can be false leads. The more knotted, the more false leads. Hence non-knotted trees gets "bonus" clues whereas knotted trees get pitfalls.

How Do You Deal with Shared Matches When You Are Related to a Match Multiple Ways?

So let's wrap up all these "solutions."

First, build out the tree of the test-taker.
  • Make it lush and green! If you don't know why your shared matches are mixed up, you'll often figure it out when you go beyond the direct ancestors and find the siblings, cousins, and in-laws. 
  • The in-laws of biological relatives are often the key to understanding the problem. 
  • That's also where all those unfamiliar surnames come from so just do the research!

Next, build trees for matches.
  • Don't just stop when you find a shared ancestor if other branches of their tree might also contain shared ancestors. 
  • Do make judgement calls like not researching a German immigrant's branch when your whole family is from Mexico.
  • Take notes of your choices in case you need to revisit.

Use the simple steps described in my free course "The Road to DNA Success."

Try and identify which of the four grandparent branches each match belongs in.
  • You likely can't do this for every single match but at least getting them to the grandparent branch will help with what you do next.

Add appropriate tools based on your goal and where the test results are.

Keep building trees for matches and refining your analysis.
  • You will use different tools based on what you are working on.

That's it. It reads pretty simply and when you are focusing on specific tasks instead of "all your matches," it feels much simpler.

Here's a little tip that might help. You could easily spend a year just focusing on building trees for matches. In the last six-months or a year, I've started finding multiple new ancestors. They are now coming much faster. It isn't that I've learned to use the tools better, it's that I have so many trees for matches. I can now fit most closer new matches into an existing tree.

I'm no longer working with three matches trying to figure out the relationship. Now I often have groups of a dozen or more matches. I might have three that have a shared ancestor (not found in my tree) and three or four with another shared ancestor (also not in my tree).

From analysis I know they belong in the same bucket and go with another group of matches. Then I'll find they all share a much more distant ancestor. I could never have identified that more distant ancestor if I'd only had three of four matches.

Even if a cluster tool showed all those matches together, that was still meaningless without having trees for them. And just the trees provided by the matches often aren't enough.

Build trees for matches. Keep building trees for matches.

Use tools to determine which branches of the trees to keep building.

Use tools to determine which matches likely have a shared ancestor and then compare the trees you've built and build trees for any matches in that group that doesn't have one.

You will build a lot of trees.

You're a genealogist, you love doing genealogy, right?

Now you've got tons of easy genealogy to do. Come back and add on more involved research, later.

Everyone needs to build trees to use DNA results. If your family tree is knotted, you need as complete trees for your matches as possible. Sometimes you will have to set aside using a promising match if they have an unknown parent or grandparent. You need lots of wide trees so you can then decide which branches to make deep.

Building trees and layering in different tools will lead to new ancestors. It's not fast but it's very possible, even if your tree is terribly knotted!

The Road to DNa Success a free online course click here to sign up

No comments

Post a Comment