Investigating a Long-Held Genealogical Theory Using DNA Evidence - Purdy

About a decade ago when I started my family history research in earnest, I was assisted by an extremely talented genealogist by the name of Karin Corbeil. I was amazed at her generosity and the amount of time and care that she invested in researching my family line. At the time, I couldn't understand why someone would volunteer to research a family that wasn't even their own and ask for nothing in return. That was before I came to know the community of genealogists and witnessed this kind of generosity over and over again.

Karin's husband descends from Nancy Purdy (1797-1878) and I descend from Daniel Purdy (1817-1897). Daniel is one of my most recent brickwalls. Karin had long theorized that both Nancy and Daniel were grandchildren of Joseph Purdy (c.1736-1818) whose family is found in 1803 in Haldimand Township, Ontario, Canada. Joseph Purdy was the son of Obadiah Purdy and Phoebe Underhill. If her theory is correct this would make her husband and I sixth cousins.

To all of our benefit, Karin has now ventured into the world of genetic genealogy and, with a mind like hers, she is extremely well suited to the endeavor. While scanning her husband's Relative Finder matches on 23andMe.com, she came across an anonymous match with familiar ancestral surnames listed (not Purdy). Due to her extensive work on the Purdy family trees, with only four surnames to go on, Karin was able to make an educated guess as to who the person was behind 23andMe's shield of anonymity! She then asked me if I also had this individual on my list of matches. I didn't, but I did find him in my paternal aunt's and uncle's Relative Finder by searching on one of the listed surnames. Since he hadn't answered her initial invitation through 23andMe, Karin tracked down an email address and sent a message to the suspected person behind the match. Sure enough, she was correct and, even better, he was willing to "share genomes" with both Karin and I to determine more details about our match.

Karin's husband and this individual, who are 4th cousins through their Purdy ancestral lines, share .35% of their DNA and were correctly predicted to be 4th cousins by 23andMe. Interestingly, this Purdy Cousin also matches my aunt and uncle, who would be his 5th cousins once removed if Karin's theory is correct, on .14% of their DNA and was predicted to be a 5th cousin. My sisters and myself, who would be his sixth cousins, do not show a match with him at all on 23andMe.

Purdy Cousin compared to me, my uncle (green) and my aunt (light blue)

We also investigated our matches on Gedmatch.com to determine if there might be some smaller matches that fell just below 23andMe's matching threshold. Since my father was tested at FTDNA instead of 23andMe, I was also able to compare his DNA to that of our Purdy Cousin using Gedmatch's tools.

On Gedmatch the following matching stretches of DNA were detected with our Purdy Cousin:

With Karin's Husband:
Chr Start Location End Location  Centimorgans (cM)  SNPs
1     88,186,177        94,337,025          5.6                   2,240
1     94,337,035       114,454,559        19.6                  7,219

With Me:
Chr Start Location  End Location Centimorgans (cM)   SNPs
1       5,314,984       7,050,643            3.6                        503
8     136,313,206    138,665,940         3.9                        554
9     127,867,126    130,999,928         4.4                        747
12    24,589,880      26,489,079          3.3                        689
12   114,422,485     116,042,182        3.4                        516

With My Uncle:
Chr Start Location   End Location Centimorgans (cM)  SNPs
1      206,237,168      215,353,110       12.3                   2,045

With My Dad:
Chr Start Location   End Location Centimorgans (cM)  SNPs
1        5,044,926        7,029,550           4.0                      624
3       29,876,878      31,804,154          3.0                      587
10     73,415,605      78,397,973          3.7                      902
16     53,357,205      55,155,137          3.4                      576
16     78,611,267      80,033,433          3.5                      571

With My Aunt:
Chr Start Location  End Location  Centimorgans (cM)  SNPs
1       206,246,175    215,354,364    12.3                       3,487
19       8,636,038        11,088,559     5.6                          809

Many of these segments are too small to be meaningful, especially the matches that I show on Chromosomes #8, 9 and 12 since they are not detected in my father's DNA. However, when looking at this data as a whole, it is apparent that there truly is a familial relationship between these individuals. Although we cannot be absolutely certain that the shared DNA that has been detected between them is from our Purdy ancestral lines, I feel comfortable that this data combined with Karin's many years of solid research shows very strong evidence in support of this conclusion. Karin writes that this is "... a huge breakthrough in the genealogical community for Purdy researchers. For many years now, I and a number of family researchers and historians have been trying to connect Nancy Purdy (wife of Harnden Eddy) to other Purdys in Ontario, Canada."  In time and with more Purdy descendants in the database, I believe that we will be able to determine beyond a reasonable doubt if this DNA is indeed inherited from our presumed common ancestor Joseph Purdy.

On a side note, I recently discovered that Karin is adopted and searching for her birth family, which is astounding considering her genealogical prowess! She was born as Carol Lee Foley on July 22, 1945 in Brooklyn, New York. If anyone has any information that might help Karin find her birth family, please contact me and I will pass it on to her or visit her new blog. I am quite confident that with her research skills and the newly added valuable tool of DNA to her toolbox, Karin will soon solve this mystery. The family that she finds will be very lucky to have her. I wish her and all adoptees success in their searches. Everyone deserves to know who they are and, in this era of personal genomics, it is finally becoming possible.

[Update - Karin Corbeil has reunited with her birth family, in part, thanks to DNA testing.]

Known Relative Studies with 23andMe: Second Cousins or Half Second Cousins? (2nd Cousins, Part Three)

In the first two installments of my second cousin studies, I have focused on my maternal side. This time I am focusing on a second cousin from my paternal side. Our shared ancestors are our great grandparents Willard Moore (1877-1934) and Blanche (Purdy) Moore (1888-1935) from Washington State. I am descended from their first son Fred and this cousin is descended from their second son Jack. I don't have as many family members tested on that side, so I have less comparisons to share, however this study brings up some interesting questions.

Fortunately, I have both of my sisters tested at 23andMe. The amount of shared DNA between us and our paternal cousin (PCousin) are 1.15%, 1.87% and 2.4% (in order of our testing). Since the expected percentage of DNA sharing is 3.125% on average with a second cousin and 1.563% with a half second cousin, I was beginning to wonder if we were really full second cousins, sharing both of our great grandparents. However, as my sisters' results came in the percentages were higher and higher. I think it is very interesting that this second cousin shares twice as much with the third sister tested as the first. The percentages are still a bit lower than expected, so more comparisons are needed to determine the exact relationship.

Notice also our paternal cousin shares on only two small places with all three of us. The other blocks of shared DNA are all randomly distributed.

Click to enlarge chart

Since I have tested both of my father's siblings, I am also able to compare them to this cousin. Their relationship is first cousins once removed with an expected percentage of DNA sharing of 6.25% on average. When my uncle's results came in, the match was only 3.33%, again consistent with a half cousin relationship. Then, against the trend, my aunt's results showed sharing of 6.68%, slightly over the expected percentage for a full first cousin once removed relationship and double that of her brother. This was a bit of a relief, but still does leave some question in my mind about whether, in reality, we share both of our great grandparents or only one. I am attempting to get a male Moore cousin descended from my grandfather's brother to test to make sure that he shares my father's and uncle's unusual Y-DNA haplogroup of I2b1.

To more clearly demonstrate the randomness of autosomal DNA inheritance, I compared this paternal cousin to each of the second cousins and their offspring. The second cousins once removed are expected to share 1.563% of their DNA on average, but, in reality, they share 1.26%, .92% and .33%. When you see this, you can begin to understand why at the next step, the third cousin level,  approximately ten percent of the time there is not enough shared DNA to detect a familial relationship.

Look at the chart comparing Second Cousin #1 and her offspring to PCousin. Do you see something that doesn't make sense?

The surprising thing is that the offspring (2nd cousin1xR #1) of the sister (2nd cousin #1) who shared the least with our paternal cousin, shared the most DNA of the three second cousins once removed. In fact, she shares more than her mother does (95 cMs vs 86 cMs) and appears to share an additional segment on Chromosome 22 with our paternal cousin that her mother does not share. Since there is definitely not any shared ancestry on 2nd cousin 1xR #1's father's side with this cousin and I cannot compare my deceased father to see if the match shows up in him, I looked at the both of her aunts to try to determine if they also shared this segment with PCousin and for some reason it just isn't showing up. 

Comparing Second Cousin1xR #1 with her aunt (Sister #2)

Sure enough, there it is on Sister #2's chart. This shows that for some unknown reason, the match did not show up in Sister #1's chart, but it must be there. Perhaps a no-call is breaking the segment into two, thus making it appear to be too small to be of significance.

[**Update - Dr. Ann Turner emailed me about this scenario writing, "...no-calls are treated as if they match, so that's not an explanation. It could be a miscall or a 'fuzzy boundary,' where the child has inherited enough alleles from the other parent to make it appear that the long consecutive run of SNPs is continuing."  She asked if I have access to the raw data for this cousin (which I do not) and further suggested, "If so, you could run David Pike's utility http://www.math.mun.ca/~dapike/FF23utils/pair-comp.php with more liberal parameters." This explanation had occurred to me, but the fact that the match was over 10cMs and appeared in both individuals of different generations led me to believe that was not a possibility. I will update if I learn anything further.]

Regardless, it is exciting to know that these shared segments come from one or both of our shared great grandparents. This realization certainly makes me feel closer to my ancestors that I have invested so much time trying to get to know through my family history research.

Willard and Blanche Moore - It's their DNA!