Some tricks for working with DNA from endogamous groups

So, most already know that endogamy makes working with autosomal DNA a bit… complicated, especially beyond the second-cousin-or-so level. Over the time I’ve spent working with DNA and analyzing my family members’ matches, I have learned a ton and picked up a few “tricks” that have come in handy from time to time. There are even some cases where being from an endogamous population can prove useful in DNA analysis (although, I will admit, it mostly muddies up everything and makes it harder, haha).

I detailed one method I have used in my last post, although it’s for a pretty niche situation and has little to do with endogamy per se. However, without using this method, I would have had no hope of figuring out anything with DNA, because traditional methods fail when all your cousins are related to you and each other hundreds of ways.

Basically, I wanted to figure out which branch of my family I had gotten a particular ethnicity from (Puerto Rican). Because this isn’t a common ethnicity to find in my family’s group, I was able to analyze my matches’ ethnicity results to work out which branch contributed that ethnicity to me. There are so few people with that ethnicity in my matches that I could see them all clustering on one line. This wouldn’t work for any other ethnicity that shows up on my test, but for an out-of-left-field one it was actually really effective. Sometimes looking at what is outside of the norm can help you reason about things in this way.

Another method I have used is to find additional cousin matches descended from certain lines on my paternal side by searching for the surname in my mom’s matches. Both of my mom’s parents’ families have been in Louisiana since it was Spain. There is no branch of her family that has not been there that long. She matches everybody. On the other hand, my dad’s family is a bit more varied and has branches that arrived in Louisiana much more recently (3/4 of his grandparents weren’t from Louisiana). So for the lines on that one grandparent who was from Louisiana’s side, I can often find additional cousins on those branches by searching in my mom’s matches. These cousins are cousins of my dad as well, but they may not match him if they are quite distant. However, chances are they will match my mom on some line, somehow, and therefore they pop up for her test and I can then know that they exist and maybe analyze further.

I was able to find additional Gautier cousins by doing this and add further data to support my conclusion that our Puerto Rican ancestor is on that line. I’m also currently attempting to do this to find more Gautiers descended from an earlier generation so I can compare, because I suspect that we may have an NPE in that line and hardly have any matches above that level using my dad or grandfather’s test. I hypothesize if more people have tested on those lines and don’t match us, they may well match my mom on some other line and be discoverable that way. I found at least one new one already. If I find many more and none of them match my paternal grandfather or have Puerto Rican ethnicity, that supports my NPE hypothesis. (Obviously this is not a conclusive method to prove anything, but it can still be useful.)

The next method I have used is related to GEDmatch. GEDmatch doesn’t make it super easy to tell which side a match is on—maternal or paternal. If all your matches are related, it’s much harder, because you can’t just peek at the matches in common to be able to tell. You will generally see the same people in shared matches whether it’s a maternal or paternal match.

The best way I have found for figuring out which side a GEDmatch match is on requires you to have the DNA of at least one close maternal or paternal relative available on GEDmatch as well—parents and grandparents are ideal, but a half sibling, aunt/uncle, or first cousin could also work. I am fortunate enough to have both my parents and my paternal grandfather’s raw DNA files available there, but before I had them, I was also able to use my mom’s uncle’s test for this in some cases. She shares 25% of her DNA with him—so about half of what she got from her mom, they share. By comparing a given match with his test, I could sometimes infer the side. If the uncle shared a significant amount of DNA (similar to what I share) with a given match, it was probably a maternal one. (And, more specifically, a maternal/maternal one.) If you have your parents DNA it’s even easier, you can just see which parent of yours shares the same segments with the match as you do. It is not sufficient just to check if they match the parent at all in endogamous populations, because chances are, they match both parents. But you can visually look at the segments to see which ones came from which side if you have at least one parent’s DNA tested. In many cases, the match is very distant for one side and closer for the other.

A (not perfect) heuristic I often use on Ancestry when trying to figure out if a match is closer to my maternal or paternal side is by looking at the matches in common and seeing if they are predominantly maternal or paternal. Often this will get me on the right track. It helps if you have Pro Tools to see the actual amount shared between matches. This is useful for those both/unassigned matches as well as for more distant matches where you can’t easily see a common ancestor. In fact, I think that Ancestry’s side labeling is still quite imperfect, because I have found several cases where a match is labeled one way or the other but I suspect a connection on the other side; I look at the shared matches and they are all on the opposite side of the label. It makes me suspect the label is wrong sometimes, particularly for more distant matches.

Overall I find that being a bit creative in how you reason about matches goes a long way when you’re from an endogamous group. The standard Leeds method-type analysis has limited usefulness for us, but we have the advantage of having, in most cases, quite well-documented ancestors, and the fact that so many of us share the same ones means lots of people have worked on your ancestors before. Finding connections is always really easy; finding evidence for the one connection you are interested in right now is much, much harder. Sometimes it makes more sense to look at things in the aggregate; how many of my known cousins on a line are matches for this person, and do they match them similarly to how I match them, or is it much more distant or much closer than the match with me? Which ethnicities do the group have in common? Can I find a common ancestor within a group of matches in common if I remove my own tree from the equation? Sometimes I can’t discern a pattern at all, but often I can.

I hope some of these tips can be useful to someone else. It’s definitely an inexact science, but I have found autosomal DNA doesn’t have to be completely useless even if all your cousins match each other. In my experience, it can help guide us to the right places to look for further evidence, at least.