Using distant DNA matches to find a close relative

+10 votes
I'm working with my mother to find out information about her father.  She knows nothing concrete other than a possible name and home state, but even those are questionable.  So we're looking at genetic information instead.  Using FTDNA and GedMatch, I've found several genetic clusters of 2nd+ cousins.  Through Ancestry, Wikitree, and other sites, I found common ancestors for each cluster.  These go back 4-6 generations, sometimes further.  My original plan was to then track the descendants of each common ancestor forward and find where all the trees intersect, hoping to eventually find his parents/grandparents.  I'm running into two problems.  First, of course, the sheer number of descendants is overwhelming, and I can't be certain I'm finding all of them.  Second, all of these genetic matches (or their ancestors) track back to West Virginia.  Small, relatively isolated population with lots of intermarriages.  So the trees often intersect multiple times (and in fact the clusters of 2nd cousins often have multiple ancestors in common, so I can't even be positive I'm tracking the right families).  

Any suggestions for how to approach this problem?  I had hoped that the Global Connections feature here might be of some use, but having played with it just a little bit it seems like it doesn't quite meet my needs.  Other thoughts, other than the brute force approach of developing complete family trees for half a dozen families from West Virginia in the late 1700s-early 1800s?

Thanks for any suggestions!
in Genealogy Help by Lisa Hazard G2G6 Pilot (222k points)

3 Answers

+11 votes
Best answer
It's an interesting problem, one that needs organization to manage all the data, and filter out the irrelevant.

My main suggestion is to "fish in every pool", applying that 2 ways.  First, you've got a good start by trying ancestry, FTDNA, and GEDmatch, but you may also want to upload to MyHeritage too, and possibly also (although I find their DNA support to be very weak, over-hyped).  What you want is to find that tester who's a closer relation, whose ancestry trail will cross yours at a more recent ancestor.  That would help you eliminate a number of descendant lines from previous generations.

Second, looking in more databases may help you find more of the right descendants.  You're looking here and on Ancestry, try also FamilySearch and  Both of them are huge and are free or allow free usage, enough to investigate families and descendants.  As you probably know, no database is perfect, and all of them (including WikiTree) have numerous profiles with no sourcing or evidence and unreliable information, but all of them do have valuable data that's different from anyone else's.

A last suggestion, to help organize it, determine which ancestors look the most likely and consider them anchor persons, and concentrate only on descendant lines that include them, checking every database for what they have on them and their descendants.  At any point, you can drop them if nothing is working, and appoint a different person as your anchor, on probation if you like.  You probably have your own methods, but I get overwhelmed at times, when there are too many choices or ways to go, and I need to find a focus to concentrate on, knowing I can change it later.
by Rob Jacobson G2G6 Pilot (128k points)
selected by Pip Sheppard
Thanks so much!  Those are great suggestions.  I've dabbled with FamilySearch a little but haven't looked at at all.  I'll get my mom to upload her DNA file to WikiTree, and MyHeritage.  (Is it possible to upload existing DNA results to Ancestry, or do you have to purchase their test?  She did hers through FTDNA but I can't see anything at Ancestry about uploading files.)
No, AncestryDNA does not take uploads, they require you to buy their test.  They do have the largest database of testers though, which makes them the largest pool to fish in, and therefore the biggest probability of having the closest related testers.

I find their DNA Circles to be surprisingly good, although they work with what you have entered into your tree, and are therefore dependent on the quality of that tree.  Since you can't enter what you don't know, DNA Circles may not work for you.
OK, thanks; we'll consider that.  One hopeful/frustrating thing:  when my mom first did FTDNA, there was one very close (1st cousin/uncle/half-sib) match, but they were anonymous and efforts to contact them were unsuccessful; the profile has now disappeared.  So on the one hand it's frustrating to know there's a very close relative out there and we can't seem to find out who they are.  On the other hand it tells us that there are close relatives out there to be found!
That was sad!  It's a common problem, many people are only interested in the ethnicity predictions, whether they can wear a kilt or not, and have no interest in the genealogical aspects, may disappear if someone contacts them.  And I should warn you that many Ancestry users are bad about that - no tree and won't respond.

Thank you, Pip!
I did notice that FamilyTreeDNA, once you've placed a match in your family tree there, will start making guesses as to if the remaining matches are maternal or paternal. If you did that, you could start ignoring the "paternal" hits.
Yep, I've seen that.  It's more conservative about matching than the "In common with" button, but still useful.
I tested at 23andme because I wanted my ethnicity but now Im hooked! (Also did the FTDNA test) I made a list of 33 of my highest percentage names from 23 and me.  I have verified only 8 connections. All of them list their gender, Only two list their last names. 17 list year of birth, 19 list location. The worst is "Brittany M -  female" . How am I supposed to connect with her?. Only 3 have responded to my request to connect. Don't mind me, Im just "whining"
Aurora, I've been at this for a while longer now and have learned a lot since this post... my best piece of advice would be to also test at Ancestry as a couple of other people here mentioned.  I did a huge amount of work connecting FTDNA matches and solving my puzzle, then after the fact I had my mom test at Ancestry and immediately found two first cousin matches.  Would have saved me a lot of time if I had just started there.  If you can't do that, sometimes you can find a link to a particular match using just tiny bits of info and some Googling.  One approach is to look at shared matches to possibly find close family members.  Another is to look for them on Facebook or elsewhere and find relatives they're connected to there.  Searching for obituaries that list the matches as relatives can also be helpful.  Good luck!
+4 votes
I just recently discovered the couple I am near certain are the parents of my paternal grandfather in a situation similar to yours except that I had no name whatsoever and only knew the birth place and date for my father.

First I'd second everything that Rob has said, all very good advice.

Some additional things that helped me in my search.

I was fortunate enough to be tracing an all male line ancestor so I was able to gain some information from my yDNA test. I would have to add that this was not quite as useful as I'd originally thought it might be, but it was definitely helpful just the same. Autosomal tests for my sister, mother and I showed two unexpected ethnicities, Irish and Scandinavian. The yDNA test made it quite clear that my paternal grandfather was of Irish paternal origin. Combined with the info from the autosomal tests I was able to deduce that he also had Scandinavian ancestors. Info from auDNA matches made it fairly clear that the Scandinavian ancestors were very recent and were a mix of Norwegian and Swedish. This was based on several matches in the 30 - 70 cM range who either had nothing but Swedish or Norwegian ancestors or whose only ancestors that appeared to be feasible common ancestors were Norwegian or Swedish.

Next I would suggest combining your matches' chromosomal match data to develop triangulation groups across multiple sites. This is relatively easy if you are handy with a spreadsheet. Basically you download the match by chromosome data from FTDNA, GEDmatch and MyHeritage (which BTW provided the match that enabled me to find my grandfather in the end). Then standardize the data by reformatting it to use the same columns for the same information. Stick it all in one spreadsheet and sort by chromosome, start position then end position. This makes it much easier to compare matches from different sites to see if they match you on the same segment as others who you believe might be connected through you maternal GF.

Next use all the available info to try to develop trees for matches of interest who don't respond to requests for info. On GEDmatch I do a user search using the email address of the match. Sometimes the match will manage multiple kits and one of the other kits will have a tree attached. After that I look at their matches to see if they have any close matches who might have a tree. If you get lucky sometimes you can find a first or second cousin who ends up having a tree and one of their GPs or GGPs shares the surname of your match, bingo. Similar tactics can be productive on MyHeritage (where you can see the shared cM between your matches and the matches they share with you). Finally search for people on Familysearch and other genealogy sites. While this typically only works well for older people it has proven very productive for me.

Beware endogamy, it is the bane of genetic genealogy (or at least one of the major ones). I try to focus more on matches with single large segments than those with multiple smaller segments. On my maternal grandmother's side (primarily French Canadian) I've found common ancestors two or three generations further back than would be suggested by the sum of their matching segments.

If you have any other siblings who will test I'd highly recommend you do so as it will significantly increase the number of matches you have to work with. My sister and I have unique matches on my paternal GFs side who are probably fourth cousins. Siblings or half siblings of your mother could also prove quite useful as well.

I'd test at Ancestry (and I suppose 23&me) if it's not a financial burden, but would second Rob's warning regarding cooperation. I found three 3rd-4th cousin matches almost certainly through my PGF at Ancestry and two of them have not responded to messages. The third replied and offered to help but has simply ignored requests for access to their private tree, frustrating indeed. But don't let it get you down, I've also come across several people who've been incredibly generous and helpful.
by Paul Chisarik G2G6 Mach 2 (24.0k points)
Thanks very much; that's helpful.  Unfortunately since we're looking for my mother's father, Y and mtDNA can't help us.  But I'm trying to use the geographic info as much as possible.  I've got clusters in particular counties in West Virginia.

I haven't systematically pooled genetic info across sites, but I'll consider that once I have more sites.  I started by just working with FTDNA and have pretty thoroughly covered that group.  We later added GEDMatch, which got me a few more people and helped plug some holes.  Today we got started with MyHeritage (and and WikiTree), and we'll see what matches we get out of those.  ( was a bit disappointing... only 118 matches, compared to ~4000 on FTDNA, and the closest was a third cousin who was pretty far down our list of FTDNA matches.  Still, there might be some gems hiding in there.)

I've gotten pretty good at finding info on people who don't have trees or even full names posted, just with their email or user ID and Google (sometimes I can find them mentioned in an obit), or as you suggested by finding relatives who do have trees posted.

Yeah, the endogamy thing is an issue.  I've found one "third cousin" for whom 4 of her 8 great grandparents are from the same family.  She could be a 5th cousin 4 times over or something.  So I've been assuming that the estimates of relatedness are probably off by at least a generation, probably more in some cases.  

My mom doesn't have any full sibs, and if there are half sibs we're not aware of them yet, so unfortunately that doesn't help.  She has a maternal second cousin we're using to at least partially screen out maternal relatives on FTDNA.  

Trying to get my mom to test on Ancestry as well.  Would there be any advantage to getting me (or my brother) tested?  I've been assuming not, because we're another generation removed from the connections we're trying to find, and since we got our genes from my mom we wouldn't really get any new genetic info.  (We already have extensive genealogical info about my dad's family, so there's no huge motivation to do it for that reason.)
If you're not looking to do any work on your Dad's side then I'd say that there's no benefit to you or your brother testing apart from the outside chance that confirming or refuting your dad's ancestors may offer some minor clues to your mom's if there is any endogamy involved. The second maternal cousin should certainly be useful though it's probably good to keep in mind that endogamy could have an impact on that as well. If someone matches your mom on two or more segments but only matches her maternal 2nd cousin on one of them there is some chance that the other segments could be from her father. But probably best to first focus on the biggest segment matches who don't match her 2nd cousin, once you've exhausted those then you can move onto the more complex ones. That and luck and patience, I searched for over two years (and I know many people have searched far longer) before that one crucial match showed up at MyHeritage.
My dad's family is from Rhode Island and Georgia/Florida so I suspect there wouldn't be much chance of endogamy between those families and my mom's dad.  My mom's mom's relatives are from Baltimore but originate in coastal Virginia, so odds of connections to West Virginia are greater there (and I've spotted some Jetts in WV trees).  So I'm watching out for that possibility.

Thanks for the reminder that this could be a long slog, but that patience pays off!  I initially naively thought this would be challenging but quick.  I had no idea...
+2 votes
I just wanted to follow up with an update.  Yesterday I got very, very lucky and am ~99% confident that I've found my mother's father.  I had gotten about 5 genetic clusters that I wanted to connect (Riley, Jarvis, Cochran, Cottrell, Sargent lines).  Started working on Jarvis because I had the most people connecting there.  Got a generation in and decided to switch to Riley.  There are lots of Jarvis/Cottrell/Cochran/Sargent connections and it was going to be hard to tell signal from noise.  Riley is more geographically distinct, so I figured if I started there I would more easily be able to spot connections to any of the others when they showed up.  The first Riley offspring I looked at married a Sinnett whose mother was a Sargent and connected directly to my Sargent line.  So I followed the Riley-Sinnett family (three sons).  The first son I looked at married a Jarvis that tracked back to my Jarvis line.  So I had three clusters linked.  Cottrell and Cochran are intertwined with Jarvis, so they're in there, too.  The Sinnett-Jarvis family has about 10 kids.  One of them has a birth date and name that almost exactly match the information my grandmother had given us (last name is different but similar), and he was in DC at the right time to have been involved with her.  Looked into it, and everything about him matches up with the info she had about him.  (Including the news articles about him; unfortunately he was not a nice person, which also fits with stuff Grandma said.)  His tree fits the genetic data; all four of his sets of great grandparents connect to genetic matches to my mom.  Though it's not impossible, I would find it very hard to believe that there's a better candidate out there.  I think I got incredibly lucky here; I thought this was going to take a lot more time and effort.

So a new question, I guess.  At what point am I confident enough to connect him to my tree?  Do I need to try to find living close cousins to do genetic testing to give us more evidence?  (He's no longer living and doesn't appear to have had other kids.)  There was no marriage, so there's unlikely to be any existing documentation.  

And thanks to all who gave me advice!  It was all very helpful.
by Lisa Hazard G2G6 Pilot (222k points)
Congratulations on your breakthrough. The other nine kids will share roughly half of their DNA with him about 35% to 65%, but not all of it (unless an identical twin). If you can find one of them, or a close descendant who is willing to test, then you could see if the DNA match was in the appropriate range. (Bear in mind you may come across the person who hid themselves previously. Contact the adoption angels on here for advice - they may be able to suggest an approach or act as a neutral broker.)

If the connection still holds up I suppose the only way to be very certain is to at significant expense test descendants of the brothers of the suspected father to eliminate them. You might get an ambiguous result though if one of the brothers shares a lot of DNA with the actual father.

Good luck

Unfortunately none of his siblings are alive, so first cousins of my mom would probably be the closest we can get.  That's at least potentially achievable, so that's part of the plan moving forward.  Adoption angels are a good idea; I hadn't thought of that.  Thanks!

Related questions

+10 votes
4 answers
182 views asked Aug 22, 2017 in The Tree House by Lisa Roberts G2G6 Mach 2 (21.8k points)
+3 votes
2 answers
+4 votes
0 answers
+5 votes
3 answers
230 views asked Mar 25, 2021 in The Tree House by Sherry Holston G2G6 Mach 1 (18.0k points)

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright