This post outlines how I recently evaluated shared segments when investigating a new DNA match. I also considered the conclusions I can draw by comparing the new match with other close relatives who have tested.
A new “Theory of Family Relativity”
MyHeritage recently ran its Theory of Family Relativity process again. This process analyzes the trees of your DNA matches and attempts to figure out what your relationship to them might be. I logged in a few days ago to see if any interesting new matches were included.
I was able to filter my match list by clicking on ‘All tree details’ and selecting the filter option ‘Has Theory of Family Relativity’.
As a European tester with lots of common names in my tree, this kind of automated process is less helpful for me than for others. In general, I find the tests I administer have either:
- Some suggested connections I was already aware of (in most cases, the connections have been inferred from my own tree!)
- Several suggested connections based on a common name and a small DNA match, with no strong geographical connection
- No suggestions at all
Have I struck gold?
However, a good genealogist leaves no stone unturned, so I always take a look anyway. This time I was in luck: the diagram excerpt below suggests that Paul Feathers (not his real name) is my 5th cousin
I feel confident about this:
- The genealogical info in the tree looks feasible; Paul’s ancestor Emily is already in my tree
- Although I don’t have many shared matches on this side of my tree to help corroborate, 22.8cMs is a reasonable amount to share with a 5th cousin.
In fact, it would have been far more likely that I shared no DNA at all with a 5th cousin. So even though this is only 22.8cMs, I feel as if I’ve hit the jackpot. It seems as if at least some of this 22.8cMs of DNA has come from my ancestors John Couzens and Sarah Cutler, born nearly 250 years ago in Hampshire, England.
It’s at moments like this that I shiver in amazement at the way genetic genealogy helps reveal things I couldn’t possibly otherwise have known.
However, I’m a born sceptic, so I’m not going to just take these 22.8cMs of DNA at face value and assume they all come from this known genealogical connection. I can examine these segments and see if they make sense alongside what I already know about my paternal chromosomes.
Investigating with chromosome mapping
Although I haven’t done much recently, I’ve developed quite a detailed chromosome map over the years, which is very helpful when new matches come in. I’m lucky enough to have both parents tested, and to have completed a visual phasing project for myself and my siblings. This means I have all of my DNA assigned at least to grandparent level, with additional detail in certain locations.
To get the segment data for the DNA that Paul and I share, I scroll to the bottom of the MyHeritage match detail page and click ‘Advanced Features’ and then ‘Download shared DNA info’. This gives me a comma-separated-values (CSV) file, which I can open in a text editor or a spreadsheet program.
As I suspected, the segments are all quite small. I know from experience that smaller 6-7cM segments can be false, but I’m going to persevere.
Painting the match
I can now return to my chromosome map at DNA Painter and click Paint a new match.
This brings up a blank form. I can copy the segment data from the CSV file and paste it into this form.
There are two buttons: ‘Preview’ and ‘Save Match Now.’ In order to evaluate the match without painting it, I click ‘Preview’. I also lower the segment threshold to 6cM in order to include the two smaller segments so that I can see them in context.
What can I conclude?
I can see immediately that one of the segments looks wrong:
- The smallest segment of 6.2cM on chromosome 18 falls on an area that my research so far has shown to consist of DNA inherited from my paternal grandfather, as opposed to my paternal grandmother, who was a descendant of John Couzens and Sarah Cutler.
- My grandfather was a German Jewish man who came to England in the early 1930s and had a completely different ancestral background from my grandmother.
- Also, this segment spans the centromere, where fewer SNPs are tested.
- I can therefore conclude that this 6.2cM segment (the one to the right of the diagram above) is probably false.
On the other hand, the other two segments, while small, could be genuine:
- These two segments on chromosome 8 both sit within areas of that chromosome that my research has shown to consist of DNA inherited from my paternal grandmother.
- I can hypothesise that the two segments reasonably close together could be evidence of one longer segment shared by an ancestor that was then broken in the intervening generations.
So, what does this mean? All I can really say is that I haven’t yet disproved that the two segments on chromosome 8 were inherited from John Couzens and Sarah Cutler.
Both segments are in the ‘danger zone’: the chance that a segment is a false match starts to rise dramatically once you get below 10cM. I would prefer it if these segments were larger, but I’m also keen to uncover evidence of a genetic link to Couzens and Cutler if it exists.
Comparing to other known relatives
The segments on chromosome 8 overlap with segments I share with some known relatives of my paternal grandmother:
- My second cousin Denise
- Mack, my father’s maternal half-brother
- My father’s maternal first cousin Chris
Working on the assumption that the segments I share with my 5th cousin Paul really did come from John Couzens and Sarah Cutler, I can infer some additional information about the segments that overlap on the same copy of chromosome 8.
Denise’s grandmother Oona was my grandmother Enid’s sister, and Chris’s father Victor was their brother. To help me consider what the possibilities are for any segments shared by their descendants, I’ll look again at the tree and try and think in terms of DNA being passed down:
On the paternal copies of their chromosomes, Victor, Oona and Enid inherited DNA from either Charles Jones or Mary Moore. On the maternal copies, they inherited DNA from either Harry Cooper or Mary Allen.
The children of Victor, Oona and Enid are first cousins to each other. The segments they share came from their parent’s maternal or paternal copy of each chromosome. From this we can conclude that each shared segment was most likely inherited from one of Charles Jones, Mary Moore, Harry Cooper or Mary Allen.
An exception: it’s possible that in certain places, Victor and/or Oona and/or Enid had recombination points close enough together that the testing company algorithm has merged segments inherited from more than one grandparent.
Returning to my paternal copy of chromosome 8, I can see that the large segment I share with Denise, granddaughter of Oona, overlaps both the ‘Couzens-Cutler’ segments I share with Paul. John Couzens and Sarah Cutler’s granddaughter was Oona’s grandmother Mary Allen. This leads me to tentatively conclude that this entire segment might have come from Mary, pictured below.
If this entire 43.3cM segment came from my 2nd-great-grandmother Mary Allen, this information could potentially help a great deal when it comes to identifying any other matches who share DNA in this position on my paternal chromosome 8.
How safe is this conclusion?
To repeat my chromosome mapping mantra: ‘always be prepared to be wrong’… I have to be realistic and see this as far from proven, particularly since the segments are small. But most of my matches are quite distant, and I can only work with what I have. So I feel I should at least take this match seriously and consider it.
The final twist
To recap, I appear to have a solid 5th cousin link to Paul Feathers via our common ancestors, my 4th-great grandparents John Couzens and Sarah Cutler.
According to MyHeritage, we also share 22.8cM of DNA. The segments we share are small, but two of them are close together. Furthermore, these overlap segments I share with my 2C and 1C1R, both of whom are also descendants of John and Sarah.
But a closer look on the match detail page at MyHeritage reveals a few problems:
- My father, my 1C1R Chris, and my second cousin Denise all have tests at MyHeritage
- They all share DNA with me in the relevant positions on chromosome 8
- But Paul does not show up as a match to any of them
- One might then speculate that Paul is a maternal DNA match, in spite of his connection via my paternal ancestors.
- But Paul does not match my mother either!
What can I make of this?
This sounds impossible, but actually it’s not that unusual. By my reckoning, one of two things are going on here:
- Paul does actually match my father, 2nd cousin and 1C1R, but due to errors in phasing and/or imputation, the match does not show up
- Paul does not match my father, 2nd cousin or 1C1R, and he also doesn’t really match me
- Instead, a mix of DNA from both my parents has led to segments that are not genuine matches
- This, combined with the genuine genealogical connection, has led to a feasible-looking Theory of Family Relativity
- This doesn’t mean Paul isn’t my 5th cousin, but means these segments of DNA are not genuinely shared with him
Which is it? I don’t know! Looking at the evidence I have:
- Some matches who triangulate with me and Paul also show up as matches for Denise and Chris on the same segment
- I therefore have a hunch that it’s a genuine match, and that we are seeing a false negative in terms of Paul not matching my father, Denise or Chris.
- With luck, some additional matches will come up who will help me to corroborate the source of this segment
In the meantime, let’s look on the bright side: I’ve found a 5th cousin!
Stop Press
I have just transferred my father’s half-brother’s DNA to MyHeritage, and he does triangulate with myself and Paul on the 9.8cM segment.
This makes me feel more confident in the tentative conclusions I drew above!
Contact info: @dnapainter / jonny@dnapainter.com