Eurogenes Biogeographic Ancestry Project

[poiuytrewq0987]

Don't know if it's just me, but it comes up as blank for me.

It's working for me.

Location of Serbs on map:

[Polako]

Why should someone with 50 percent "North Euro" having no "North Atlantic" or vice versa?

It's because the main 3 North European components are usually within 0.025 Fst of each other, and the North Euro and North Atlantic usually within 0.02 Fst. So only a handful of alleles are used to differentiate them. That's really finescale stuff.

Better sampling might fix the problem, and probably so would extra markers. But 200,000 unlinked SNPs is all I can get from the 23andME data, and that's pretty much what my laptop can handle anyway.

A STRUCTURE-like algorithm would also probably help eliminate the extremes, but it would take days to complete each run. ADMIXTURE is much faster, but more choppy.

In the end, it's best to remember that we're looking at ancestral probabilities here, and not straight up genetic structure. So these results will place us into our ethnic groups of origin, and we're likely to get some very strong hints about the type and maybe even amount of admixture from other parts of Europe. But many people won't get a set of scores that truly reflect their genetic structure across the 22 autosomes. For that we'd need something like the Ancestry Painting, with extremely well chosen and comprehensive reference samples.

[Agrippa]

Don't know if it's just me, but it comes up as blank for me.

That's because it is so huge. My computer has problems with it too, better bring it into bitmap and edit it there, probably reducing size too, because otherwise there is not too much of an overview.

The positioning makes sense it seems.

It's because the main 3 North European components are usually within 0.025 Fst of each other, and the North Euro and North Atlantic usually within 0.02 Fst. So only a handful of alleles are used to differentiate them. That's really finescale stuff.

I just wonder if no other component popped up for Europe, since this one seems to be not informative enough?

I mean, if we would have siblings in this run, I wouldn't wonder if they would come up with totally different results for their Northern components in particular, probably changed "all or nothing" results to say it that way.

Better sampling might fix the problem, and probably so would extra markers. But 200,000 unlinked SNPs is all I can get from the 23andME data, and that's pretty much what my laptop can handle anyway.

Of course, I realise the technical limitations.

A STRUCTURE-like algorithm would also probably help eliminate the extremes, but it would take days to complete each run. ADMIXTURE is much faster, but more choppy.

In the end, it's best to remember that we're looking at ancestral probabilities here, and not straight up genetic structure. So these results will place us into our ethnic groups of origin, and we're likely to get some very strong hints about the type and maybe even amount of admixture from other parts of Europe. But many people won't get a set of scores that truly reflect their genetic structure across the 22 autosomes. For that we'd need something like the Ancestry Painting, with extremely well chosen and comprehensive reference samples.

The future...


I just wanted to point out the relative character of the differentation, because some people just over-interpret it, when in reality the distribution is often too much of a random thing if looking at the samples as a whole - which is easier with graphics.

The final positioning is still good enough for most, so it is informative overall.

Thanks for your answer and efforts, very much appreciated, you're doing a great job!

Every small failure might lead to a future improvement - that's my credo for such things

[Frederick]

I quickly made some crude outlines around regions/countries and tried to mark people in wich those components peak.
Picture is also smaller
Somehow, I wasnt able to find the FIN on the map...

[Scrapple]

High Balto-Slavic ... sure you have no such ancestry? Not listed in your profile.

I haven't found it yet. Polako is an RF match of mine and he has high Baltic and I do have a 17cM 3 GP southern Poland 1 GP Lithuanian match but don't have a paper trail yet so don't know what to put down. So it is real.

All the others in my profile I can document.

Edit: I am dumb sometimes, some of my German came over in the 19th century was from the area that was known as Prussia.

[Frederick]

Polako is an RF match of mine

Funny. I have him too.
I share 5.1cM with him.

If I remember it right, Polako has, besides his Polish anchestry, "Baltic States" and "Austrian/Hungaran Empire" lineages that he knows of.

[Sahson]

I finished up the RMSD calculator for the EU7c run this morning. Here is a link to the new spreadsheet. I used OpenOffice again because Excel was being a pain.

Looks like I'm Scandian once again.



My closest 10 groups are:

DK 4.66
SE 7.06
NO 7.14
US 9.65
Kent 10.70
NL 11.91
DE 11.91
AT 16.88 (Austria?)
CH 17.02 (Switzerland?)
Cornwall 17.14

I've tried pissing about with your spreadsheet, and can't get that chart thingy, can you do one for me please AU5.

Here I am in Pie format, everyone likes pie.

[poiuytrewq0987]

I quickly made some crude outlines around regions/countries and tried to mark people in wich those components peak.
Picture is also smaller
Somehow, I wasnt able to find the FIN on the map...

Estonians clustering with Slavic Poles, oh the horror!

[Frederick]

Estonians clustering with Slavic Poles, oh the horror!

Those (member) Estonians are strange anyways.
They totaly stuck out with strange results in the several past runs.

I recall a run, in wich these 2 Estonians made Estonia extremely "North Atlantic" (not this time. Seems this time the same signal was interpreted as "Northern European") And this "Northernm European" pulls them to the left, into Poland.

Also, in this run, they have huge Balto-Slavic components and tiny East-Euro Finnic one.

From people who speak a Uralic language, I would expect far higher Finnic results.

Ok... its strange that TWO Estonians, BOTH show strange results, but lets wait for more Estonians.

[poiuytrewq0987]

Those (member) Estonians are strange anyways.
They totaly stuck out with strange results in the several past runs.

I recall a run, in wich these 2 Estonians made Estonia extremely "North Atlantic" (not this time. Seems this time the same signal was interpreted as "Northern European") And this "Northernm European" pulls them to the left, into Poland.

Also, in this run, they have huge Balto-Slavic components and tiny East-Euro Finnic one.

From people who speak a Uralic language, I would expect far higher Finnic results.

Ok... its strange that TWO Estonians, BOTH show strange results, but lets wait for more Estonians.

Maybe they're not that Finnic after all? I know for a fact that Karl carries R1a haplogroup...