Janice was searching for her biological father. Her Ancestry DNA test had found what looked to be an aunt or half aunt, “ Sally,” and also a half uncle, “Trevor,” who did not match each other, both about 20 years older than Janice. So likely each of them was related to one parent of the unknown father. This could be easy! But wait…
Notice the difference in matching DNA and that they are both listed as 1st-2nd cousins. Yet both are in the range for half aunt/uncle (500-1446 cM). Although other relationships, like first cousin were possible, they did not fit the known facts or the matching to common relatives. Trevor’s second cousins shared with Janice were all showing as third or fourth cousins to her, suggesting one generation of difference. The same was true for Sally’s closer cousins: her firsts were coming up as Janice’s seconds.
Trevor, born 1945, had just discovered via his Ancestry test that the father who had raised him was not his biological dad. His dad was actually his stepfather; he had adopted him when marrying Trevor’s mother. Sadly she was no longer available to tell Trevor who his birth father really was and the father of record did not fit the DNA evidence.
Sally knew that her mother had given birth to a boy in 1944 that had been adopted out. This child was not the son of the man she later married; his father was unknown. Sally had tested her DNA hoping to find her half brother and was excited to have found his daughter!
The problem here is that since Janice’s father is that adopted half brother, there may be no way to track him down. The best I could do for Janice was to find Trevor’s dad and Janice’s paternal grandfather via DNA. Since he had deep American roots it took less than four hours! That is of my time, but actually a day and a half of elapsed time.
Here is how it was done.
First, taking a look at the matches Trevor and Janice shared, there were seven third cousin matches, only one with a good tree (thank you Paula). Of the 275 common fourth cousins, the highest one had enough of a tree for Sally to have already found a common ancestral couple with Paula, surname Collins (all names in this article are fake for privacy). Note that the large number of fourth cousins made colonial ancestry very likely.
I went through all those common cousins and starred them while watching TV. Actually, once I got down to 30cM I started starring only the ones with trees. My fingers were tired … The reason to star the common relatives was to make it easier to use various tools on just the shared matches. I had Janice run a cluster report from Genetic Affairs for just the starred matches. Cluster One was clearly the Collins group but the next few clusters had no good trees nor surnames that stood out.
Next I used GWorks to make a database at DNAgedcom of the trees of Janice’s and Trevor’s common 3rd and 4th cousin matches. (click here for my blog post on how to use just the shared matches). Sadly most of these matches had either no trees or tiny trees, which meant that no other clear cut ancestral couples turned up. There were no ancestors in the database more than twice.
One possible next step would be to use pedigree thief and my DNA2ged tool to get more trees from the Ancestry unlinked trees to add to GWorks (click here for that post). For some cases I go to the MyHeritage and FamilyTreeDNA trees to get even more trees for the database.
However in this case, I noticed one unusual surname, Siberling, high up in the common surnames list, so I decided to build down from the common ancestral couple Sally had found, keeping an eye out for that name. I did this tree building as unconnected branches in Janice’s tree so that it would be easy to connect this information back in later, if a good candidate was found (and again while watching TV – PBS newshour). By the way, DNA2tree, the new tool for iPads and iPhones, can also work with just the starred, aka the favorites list, so might have been a faster way to build those trees.
The technique I use when building down from just one common couple is to check the surname of each child’s spouse against the GWorks frequent surname list. There were six Collins children, a few of whom had spouses with surnames in the list. I built those trees down but soon the surnames petered out. Finally at the fourth son of the sixth child, I found a Collins who had married a Siberling and had even died in the right state. He had only one son who was the right age and in the right place at the right time! Eureka! Not surprisingly this WWII vet was no longer living. Lets call him Roy.
Next I built Roy’s pedigree tree back to about 1800. Then I gave him a fake brother with the same parents and made this fake brother Janice’s dad in her tree (her DNA was already attached to her tree which had her mother’s family).
By the next day, the ThruLines common ancestor hints on Ancestry had come in. As had Trevor’s invitation to me to see his DNA results. Now I checked the numbers for Janice and Trevor on the newly found common ancestor relationships. It all fit. Plus there were good matches on Roy’s mother’s line and his paternal grandmother’s line. This is as proven as it gets without testing one of Roy’s children. [UPDATE 2020: Roy’s son tested and this is now proven]
On a side note, Janice and Sally had been crossing paths all their lives as they moved to different areas of the country. They are now 10 minutes from each other. Sally shared this with me “I own a dog who snaps at or bites everyone. EVERYONE. Janice? He leaped up and hugged tightly around the waist, tail wagging, looking into her eyes the second time he met her. The first time? I looked down to see Janice was petting him and he was loving it! THAT never happened before and he is 10. … Perfectly comfortable with her. Old friends.”
UPDATE June-2019: Janice’s father has been found using old fashioned detective work. There is a P.I., John Apel, with the birth index for Kansas City, MO and much expertise. He narrowed it down to 2 people and one of them is the guy. Sally is now in touch with her half brother! And he is getting used to having a daughter he did not know of.
Would it be an idea to only perform AutoCluster analysis for one or a couple of matches using their shared matches as matches as well? I guess it’s basically the same as starring only that interesting match or matches and use the extended clustering? The combination of starred matches and extended clustering can be quite useful I think, would that have worked?
EJ that would be a wonderful new feature! To be able to cluster only the matches in common with another specified match
I think you can already perform this feature using the existing options on Genetic Affairs. First, star the match (or matches) you are interested in. Next, enable on the extended clustering. Run the analysis. I am thinking, perhaps it needs multiple rounds? So that it uses the shared matches of the first match as matches, and then the shared matches of these shared matches as matches etc etc…
Kitty, What a great story and use of all these new and exciting tools. Congrats on the findings. Paul
This appears to be a perfect approach to finding a parent of an adopted person. I have a somewhat similar problem in that I have a great grandfather (NPE) who was not married to my great grandmother. Would I do basically the same steps to try to find him? I have worked on this for YEARS and have come up with nothing. I think maybe some of the new apps and programs that you have used here MIGHT help. Please comment as I am sure others are in this same predicament. Thanks in advance for any help.
I wrote up this case because it was such a good example of using all the latest tools.
For an unknown grandfather I would suggest:
1) get a male line descendant to do a Y37 marker test. If there are good results suggesting a surname …
2) get several 2nd cousins descended from the NPE to autosomal test at Ancestry. Yourself and 1 other at a minimum.
3) build the NPE’s wife’s pedigree tree back to at least 1800
4) Use the techniques above on the shared matches but see if you can eliminate the ones from the NPE wife’s side
5) You may need to go back to segment data and other techniques if clustering and GWorks do not give you enough data. Classes at DNAadoption can teach you how.
Thank you Kitty. I have no Y DNA to test. Have searched diligently for a male in this line to test. No males left. I don’t even have my father’s autosomal DNA to test, but I did send some artifacts to totheletterdna.com and hope maybe to get something there. I will follow your advice otherwise though and see what I can come up with. Thank you so very much.
For autosomal, woman are fine, get a 2nd cousin or two tested
Great results and very satisfying!
I do think the four hours is misleading and makes people think DNA mysteries can be solved much faster. You had to wait overnight for results to be processed, okay a technicality.
There also didn’t appear to be any issues with downloading data from sites, uploading to tools, or getting the tools to run, all things that are happening more frequently as new versions are being pushed out and more users are taxing the systems. I’d suggest a disclaimer “If everything runs perfectly….”
Very good point. I amended it to ” it took less than four hours! That is of my time, but actually one day of elapsed time.”
I always have other tasks I can do while gathers and uploads are at work so I do not count that time …
The reason it was so quick is that “Sally,” the incredibly helpful half aunt, had done so much already, namely identifying a common ancestral couple from the mid 1800s. Plus one parent of the father, Sally’s mother was known.
Searching for one parent with deep American roots can be very quick. The first time I used clustering, I identified an unknown father’s parents in less than 2 hours!
So no, I do not think it is misleading. With deep American roots and many good matches it can be fast.
Do you have any blog posts or resources for ‘half’ relationships. Is it assumed that most relationships are ‘half’ if you are adopted?
Id like to understand the challenge of trying to find a birthparent or other biological family members using mostly half relationships.
Hi Cindy –
No most relationships for adoptees are not halves, only the ones in your own generation and below. For example, a brother of your bio father would be a full uncle and his child a full cousin. However children of a bio parent are half siblings and their children are half nieces and nephews.
Use this calculator: https://dnapainter.com/tools/sharedcmv4
as for posts, try this one:
Yes, as Kitty says, an adoptee’s siblings would be half siblings, if they only shared one parent.
I have that, as well as my maternal grandmother was the only child of her father and mother. Her father married another woman, and her mother another man, and all of those descendants are half relationships to me and my grandmother’s descendants.
Thrulines is actually showing those relationships correctly!
Very helpful. Thank you!
So, in this case, I found a female match (18 years older than me) whose 87 yo father is Adopted. She matches me with 781 CM / 10.5% / 20 segments and 8995 SNPs on the X (23andme). The adopted father is being tested, so I hope his test results get posted within the next month or so. My assumption is that he is my half-uncle and she is my half-first-cousin —— unless she is a half-aunt considering her age.
Since he is adopted
That is very high for a half first cousin but not impossible, see the calculator: https://dnapainter.com/tools/sharedcmv4
So that 87 year old could not be your half sibiling? (more likely for those numbers but the age probably does not work).
Hopefully his test will clarify this.
Do you have any older family members you can ask about a child given up for adoption? Can you test a paternal and maternal first cousin and see how they compare?
Im adopted, 49 years old. The match is 67. She could also be a half-aunt based on her age. I assume a half relationship either way since her father (87 yo) is also adopted. He could either be an (half?) uncle or a great uncle based on his age. Another 2nd cousin match (not adopted) is trying to get an Aunt tested who would likely be a (half) sister to the 87yo male and maybe a sister to my dad who I suspect is also in his 70s. However the 77yo aunt is reluctant to test.
I found a paternal 1/2 brother (also adopted) on Ancestry and trying to get him to test on 23andMe. He has uploaded to GEDmatch.
It might be easier to get your 23andme matches on GEDmatch too in order to compare to that paternal half brother. Let me know when the adoptee Dad’s test results come in…
Her adopted dad’s results came back (fast!). I hope the following makes sense:
Ill call the untested suspected grandmother ‘G’ (1908), my suspected uncle ‘U’ (1932), his verified daughter ‘1C’ (1952), and our mutual cousin ‘2C’ (1963).
There is some redundancy below but trying to make it easier to see the relationships:
U: I share 15.9% (1187) with U, while 1C shares 49.9% with U (her dad), and 2C shares 10% with U. 23nMe is calling U my 1st cousin but he is 38 yrs older than me. I ran a WATO with everyone including the untested ‘G’randmother and it preferred U as my uncle as opposed to my grand-uncle.
1C: I share 10.5% (781) with 1C, while U shares 49.9% with 1C (his daughter), and 2C shares 6.79% with 1C. I also put 1C’s sister in the WATO (she tested at Ancestry with 366cm shared with me). Interesting that she shares half as much DNA with me as her sister. Could this mean they are 1/2 sisters or is it just the randomness of the DNA?
2C: I share 5.21% (388) with 2C, while U shares 10% with 2C, and 1C shares 6.79% with 2C. We are pretty sure that G is 2C’s Aunt (2C’s mom’s sister), making U as 2C’s first cousin, and 1C as 2C’s second cousin.
Perhaps a bit much for a blog comment! I prefer comments that are educational for everyone.
Regardless, I will get back to you but not until Monday and via email
Completely understandable. Thank you.
I can easily keep my questions more universal in nature. Ive learned quite a bit from your blog and your responses, and therefore really appreciate the educational aspect of you volunteering your expertise to my fellow blog followers.
This sounds like a perfect case to set up in the What are the Odds tool to see what it says is most likely
P.S. emailed you a screenshot of my WATO.
Don’t be too hasty in throwing out a possibility due to age differences.
My half brother I found in January is 81, I am 63, and have just found a half sister. She is 78.
Adoptions can really throw basis premises off!
Thanks Brian –
Ill definitely keep that in mind as more people test. Such a mystery!