How Much of These Autosomal Results are Noise

[david64]

I have had my father and maternal aunt autosomal tested and merged the results to give my ethnic origins:

77 UK
6.5 Indian
4 European Jewish
3 Armenian
2.5 West and Central Europe
2.5 Eastern Middle East
1.5 Eastern Euro
1.5 Persian
1 Finnish

* I've modded the ethnicity to be specific when the paper trail points to more specific origins. These can be 2X for actual occurrence in my father and aunt.

My paper trial suggests that the UK allotment is more or less spot on. However, the rest I am a bit suspicious of and think that ethnicities are being confused. I have a considerable paper trial for Armenian ancestry going back as far as c. 1620 on one line. 6% should be the bare minimum for Armenian, but FTDNA give half that. And the Indian, I think, has either be grossly overstated or is completely erroneous. This would mean I have a great-great-grandparent who is 100% Indian, and the photographic record doesn't suggest anything of the sort.

So, my question is, how much credence would you place on these low incidence results? Keeping in mind that the incidence for Indian was 13% and 8% for Jewish etc.

Also, how deep are these ethnicities meant to refer to? My father comes out as 5% West and Central Europe, but I doubt there is any such connection in a near-time frame. I have once ancestor who was a Hugenot, obviously born over 400 years ago; and there is a family I am descendant of that there is an undocumented rumor that they are of ancient Flemish origin.

[davidft]

When you can get full siblings who test and show different "autosomal" makeups I think that tells us not to read too much into them. I think you are reading too much into your results.

As I understand it this area of genealogical DNA testing is in its infancy and relies a lot on assumptions and very small reference groups to assign these "ethnicities"

Someone reported recently that ftDNA plan to update how their autosomal analysis reports "ethnicities" later this year. When they do I would not be surprised if your results see some significant changes. Why do I say this ? I say this because the results you get from these autosomal analyses depend very much on the assumptions you make and the reference group you use. This is why you can get a whole host of different analyses on Gedmatch.com using the same data

[hurworth]

I haven't used Ancestry (FtDNA here) but I doubt that even a person who thinks their ancestry is 100% Armenian would come up as 100% Armenian.  Armenia was at a crossroads for migration. 

You say that you've modified the terms?   Does Ancestry use the terms Armenian, Persian etc.

Yes, they ARE deep. 

Gedmatch has some tools for looking at this, and may come up with some different results.

[david64]

Thanks for your replies.

I tested with FTDNA, not Ancestry. I tested a few years ago and at that time my aunt's test came out as saying 26% Indian; and now they have been updated to what appears to be more accurate, but still questionable.

I am glad to hear that FTDNA may be planning an update. I saw somewhere on their site some time ago that their samples for some regions, including Armenia, were very small; in the order of 16 or so.

The comment on the movement of people in Armenia, and other areas of the world is a good one. I'd not thought about that. I have a chart of the various y haplogroups that occur in European countries. From memory R occurs 82% in Wales; 79% in Ireland; 72% in Scotland and 67% in England. The nearest prevelancy was 56% of one haplogroup in Poland. Many places in Europe have no real prevelancy like the Balkans. With such mixture, it is obvious to see how it could be difficult to get genuinely accurate results.

In terms of Armenia, I think it is difficult to establish what an Armenian is. You get Russ-types, Turkic-types, Grecco-types and Persian-types. Though this study shows they are less diverse than many places in Europe:

http://www.ucl.ac.uk/tcga/tcgapdf/Weale-HG-01-Armenia.pdf

Not far off the distinction of English.

The Y distribution:

http://www.sharecsv.com/s/7832abcb1c97ea350c04b9c70f35f85f/haplo.csv

[Nancy2]

My understanding of autosomnal dna is that although we get this dna from both parents, ourbrothers and sisters do not necessarily get the same.  Each time a child is born the dna for that child is taken out of the combined parents gene pool, but is different each time which would explain how sisters and brothers are so different.  I'm hoping to get some of my brothers and sisters to do dna tests to check this out.