Researching a Trace Ancestry/Ethnicity?

[M_ONeill]

A few days ago I got my results back from AncestryDNA and one small thing that surprised me was that doing the origin trick showed me that I have one tiny segment (0.05%) that Ancestry has labelled internally as 'Jewish Peoples of Europe', though shows up as 'unassigned' in my official results due to falling below the percentage threshold.

Now, I'm fully aware that such an incredibly miniscule segment could very well be noise, or a mis-reading, but as there's some amount of circumstantial evidence in my own research (and that of other people I'm working with), that suggests it might be legitimate, I'm at least looking into the possibility.

Just as an example, I have one match (12cm) whose tree contains an English family line with a distinct surname I've been researching quite extensively in Ireland. Their tree goes back to an individual b. c1608 who they claim adopted the surname and was originally a member of the Sephardic 'Crypto-Jewish' community of converso emigrants from Portugal (the strong Portuguese connection is another thing of note in the research regarding the Irish family). They sadly don't have an attached source for this claim, so I don't know if it's a family story or an attested piece of history. I'll be contacting said match in a couple of days to see if they're up for sharing info.

When looking at where I match with this person in a chromosome browser, they don't share this particular segment of unassigned DNA in question - but someone who is a shared match between the two of us does. Going through their shared matches finds more people who share the same segment; I have at least ten after a day's searching. Said segments often even sharing the same start and end points. GEDmatch usually puts the estimated most recent common ancestor about seven generations out.

Now as someone new to diving this deeply into DNA I want to ask: am I potentially on to something meaningful here, or is it a wild goose chase? Is finding so many people with the same segment of shared DNA something of note, or with such a small segment could it just be random chance at work?

Thanks in advance for your answers!

[phil57]

I am on my phone, so can't send a lengthy reply. Someone else will no doubt help. A 12cM match on its own can also have a relatively high chance of being false. Only at about 20cM or more can you reliably consider match lengths to be 100% accurate.

If you and the 12cM match have shared matches at higher match lengths then there may be something in it. You would need to investigate and establish those relationships with documented paper based research to be certain.

Treat ethnicity estimates with a large dose of scepticism. They are the least reliable aspect of DNA testing. I don't use them for research and I wouldn't even consider looking into an ethnicity estimate of only a couple of percent or less.

As an example of how autosomal matches become increasingly more likely to be identical by chance at lower match lengths, I have experimented with reducing the lowest match length to 3cM on GEDmatch and picking other tests for comparison at random. I found a "match" to every single individual that I ran a comparison against, often across several segments at small match lengths.

I have to say that I wouldn't start my DNA research by looking at lower matches. I would start with the highest, plotting them in your tree, and work your way down to the lower ones if necessary. I only include shorter matches if I need them to corroborate my paper research in the absence of, or in addition to longer matches.

Before starting on investigating DNA matches, if you haven't done it already, I would also recommend researching your tree so that it is as deep and wide as possible. E.g. don't just restrict it to your direct ancestral lines, but also research the siblings of your ancestors at each generation, their marriages and children, and bring each of those forward to the present as far as you can. It will help tremendously with identifying matches, as you will often recognise the name of your match - often totally unrelated to your direct family - or the name of one of their ancestors within a few generations back.

Good luck with your research.

[Biggles50]

Totally agree with Phil.

If you do your research on how the likes of Ancestry calculates Ethnicity you should then deduce what is presented is an Estimate and very Questionable given the small database

[4b2]

Given it's Ireland and such a tale, it is likely to be via word of mouth. Useful, but these details can get twisted. As an example, I was told a bit of genealogical details of my family when I was young. I made the diligence of recording this when I was in my teens. If I was today recording this information, I would have noted that a certain gg-grandfather died when falling into a mill. But it was actually my other gg-grandfather. As other tales, my grandmother told me that us her parents were Armenian refugees from the genocide, which was not true. While another family member had it that a certain family lived at a certain home in the lat 17th century, which turned out to be true.

If you can find a direct male descendant of this family, you could get them y tested ($89). That should probably give enough details to show if they are likely Jewish, but there is also the issue of infidelity. If you consider your seven generation tree of 128 ancestors, the chances are that about 1-5 are born of infidelity.

[M_ONeill]

Thanks for the replies, all! I agree entirely regarding not starting off with smaller DNA matches; I should note that this is most certainly a sidebar to my work on the main portion of my tree. It's only because it possibly intersects with another line of more solid research that I'm looking into it at all.

Looking further into my 12cM Ancestry match (we'll call him 'A'), GEDmatch puts our match slightly higher, but not by much: 13.7cM. I have 8 shared GED matches with A. The highest of these matches is a kit we'll call 'B'. The one-to-one tool shows me and B share 17.8cM, with A and B sharing 38.7cM, largest segment 28.7. There appear to be no matching segments that overlap between all three of us, however.

Back on Ancestry, looking through my matches, I have found two other people who appear to descend from the same family line as A. These three people all claim two shared ancestors born in the 1720s, but from there on back it gets a little messy. The trees don't agree on who the husband's father was. A claims one name, linking back to the potential Sephardic ancestor in their tree, but looking at the ages in said tree, the named ancestor would have been 14 when they fathered the son. The other matches claim another name, from another part of England, but going by the ages in that tree, this man would have been  75(!) when he fathered the son!

Looking at the above, I'm starting to think that my connection to A isn't a false match - but I think I'm going to need to do some parallel research on the trees involved before any conclusions are drawn!

[M_ONeill]

Ah, 4b2, we crossed replies!

It's funny that you should mention the Y-DNA test, as it was the results of a very distant cousin (confirmed through research and DNA matches) of mine's Y-DNA test that kicked off the research in question. Namely with two cases of this rather specific surname appearing amid a block of the other, more expected surnames from a common ancestor.

[4b2]

Looking at the above, I'm starting to think that my connection to A isn't a false match - but I think I'm going to need to do some parallel research on the trees involved before any conclusions are drawn!

If a low cM match has shared matches with other people from a line, I consider it highly likely it's not a false match. If a low cM match has no shared matches, I don't pay much attention to it.

It would be good if Ancestry lowered or removed the 20cM limit for shared matches.

[Petros]

If a low cM match has shared matches with other people from a line, I consider it highly likely it's not a false match. If a low cM match has no shared matches, I don't pay much attention to it.

It would be good if Ancestry lowered or removed the 20cM limit for shared matches.

Indeed, since they started giving new matches as a group I seem to get large numbers of new matches in the 10-20 cM range who have no shared matches, and generally no trees of any significance

[melba_schmelba]

A few days ago I got my results back from AncestryDNA and one small thing that surprised me was that doing the origin trick showed me that I have one tiny segment (0.05%) that Ancestry has labelled internally as 'Jewish Peoples of Europe', though shows up as 'unassigned' in my official results due to falling below the percentage threshold.

Jewish Peoples of Europe would usually refer to Ashkenazi Jews. Sephardic Jews have a much more complicated DNA makeup and are more genetically diverse, having more often inbred with local populations where they have found themselves, whereas Ashkenazi Jews are a very homogenous population that have intermarried over many many centuries mainly in central and eastern Europe. So even if that very small segment were accurate, it would more likely refer to a distant Ashkenazi Jewish ancestor rather than a Sephardic one. Sephardic Jews will often show large proportions of North African, Italian, Middle Eastern and Balkan DNA, with smaller amounts of West Asian or Sub Saharan African.