Search This Blog


Tuesday, May 29, 2012

A genetic map of West Eurasia with a difference (aka. the ancestral dichotomy in our genomes)

What we have here is another genetic map of West Eurasia (Europe, West Asia and North Africa), using a new program and a few new samples. I usually employ the PLINK MDS utility to run such plots (see here), but this time I chose another algorithm which has just appeared online, called SPA. There are obviously some major differences between the SPA and PLINK maps, and the main one is that each subject appears twice on the SPA map.

That's because SPA offers a mode in which it assumes that the samples are of mixed ancestry, and divides their genomes into halves (represented by two sets of PC coordinates). One way to look at it is that each half represents a parent, but that would only be accurate for people who really are of mixed origin, with their parents coming from two clearly distinct groups in terms of genetic structure. So it's probably better to say that SPA turns each of the genomes into two of the most highly differentiated versions possible, across the first two dimensions. Also, as far as I can see, the program can come up with different solutions in different runs for the same person. Some of these solutions look like they're based on very ancient admixtures, and others more recent. In any case, I think this is an excellent way to show the dichotomy in our genetic data, even if we're not mixed.

The data sheet can be downloaded here, and run with programs like Gnuplot to make your own genetic maps. The map below was produced by me with Gnuplot, and then spruced up with color by Eurogenes project member FR7. A singe location version of the same plot, minus the color, can be seen here.


Wen-Yun Yang et al., A model-based approach for analysis of spatial structure in genetic data, Nature Genetics 44, 725–731 (2012), doi:10.1038/ng.2285


Eric said...

Without including other africans, the North African samples become pretty much useless, and behave strangely, since north-africans are about 14% subsharan and 10% east-african.

anthrospain said...

Hey, could you also create the normal MDS, so we can see the new samples, in their normal positions ?

Eochaidh said...

This is very interesting. I found my two plots right away and they aren't far from each other at all. One is up and to the left (southeast?) and oddly enough is in between a French-Canadian and a SW Scot. My mom was half French-Canadian and half Scots-Irish. The other plot is surrounded by Cornish testers, which I test closest to overall. I wonder if my dad's being from Co. Wexford has something to do with that. Wexford is just across the Irish Sea from Wales and Cornwall.
I can't open the spreadsheet.
Thanks David! Miles Kehoe

pconroy said...


The "spasingle.png" can't display, with error message "Contains errors"

Also, my father's (IE6) 2 points on the map are spread apart, whereas my mother's 2 points are very close, almost touching - does this mean that my mother's parents were more homogeneous or something?

Davidski said...

^ Paul, the SPA single ancestry map works for me here on two computers.

And yes, having two halves close together suggests that both sets of chromosomes are very similar in terms of genetic structure. So that might mean the person in question comes from a homogenous group. But that's not always the case, based on some of the results I've seen.

Mark said...

Great concept on splitting up ancestry's. Thanks David!

I need some help finding my other plot (US34). I have one in the Isle section (middle Right) next to DE19, UK28, Orcadian and Cornwall but danged if I cannot find my other one. I must be blind.
US34 US34 -0.411219529 -0.018891533
US34 US34 -0.662068332 -0.117145337

Thanks in advance.

Mark said...

Ha... I see my mistake. I downloaded the single version first and not the dual version JPG as shown in the Blog. After downloading the dual, I found both.... Daaaaaah


Davidski said...


Your second half is just southeast of the main British cluster, and sitting next to DE15, AT1, UK27, NL4, BG, ES and CEU.

Mark said...

Thanks so very much David for all your work here and with this info I now have a solid picture of my genome.

So looking at the Colored Dual SPA Chart, I (US34) have a genome split on directly opposite sides of the main British Cluster Upper left and lower right.

I have quit a few 3rd and 4th suggested paternal cousins who are from England and Wales or were born there suggesting the latest paternal admixture is Western England and Wales. Paternal only match status is determined from zero matches with any of my six half maternal siblings which I use a proxy for my late untested mother.

Interesting using 23andme those with high number of matches 900 or more, have a defined US Colonial ancestry, I have over 1020 with a cluster of Southeastern USA cousins with Virginia report as the furthest north identified cousins mostly from Tenn and Alabama. My four half maternal German paternal brothers range from only 500 to 600 matches. Doug McDonald commented my overall admixture had a far northern Western basis. Via the Magnus Ducatus Lituaniae Project, I have a ADMIXTURE analysis showing my apparent maternal side is Southeastern European even thought my maternal GGGrandmother was born in WestPrussia (Poland) in 1864. My half sibling are plotting NW continental Dutch as per my maternal GGGrandfather ancestry which appears to have dominated the majority admixture in my male siblings.

This interesting fact is I also have significant YDNA STR matches at 67 and 111 markers with an English/Scot surname Cook (brickwall back five generations to Georgia USA but most generations are in Northern Alabama) with a GD3/67 and GD5/111, a match to a surname of Watterson from the Isle of Man at GD6/67 and 8/111 (His family has been on the IOM for as far back as they know, and earliest Gov records show surname back to 1417), and a Scot named Ross with paternal line back to Highlands of Scotland 1750's with a GD9/67 and GD13/111. Another Curry at GD9/67 along with many other Curry’s at 25 and 37 markers all with the same off-modal STR allele values as well. All within a genealogical time frame using FtDNA’s TIP report all at 111 markers using 30 years per generation, my Cook match is at 270 ybp 68% probability (a Sigma 1), Watterson at around 390 ybp (Sigma 1), Ross runs around 630 ybp (Sigma1).

My belief is that my paternal original has a very deep ancestry from Munster Irish- O'Curry's 500AD, on to western and northern Scotland, on to IOM and then to Western Wales and England. This is shown plainly using Fluxus networking with my cluster of 25, 37 and 67 marker haplotypes showing a strong branching to my lineage based on 10 off-modal common STR mutations starting with from Haplogroup R-L21

This combined data confirms I have very strong Northwestern Atlantic Insular paternal linage.


Mark said...

Some posted a Dual Plot with out color and marked the country areas by title if someone needs one.

Joao Bessa Santos said...

Great job David :)

Joao Bessa Santos said...

Great job David. Thank you for your hard work :)

Joao Bessa Santos

princenuadha said...

Poland vs Greece!!! You and dienekes need to make a bet : )

Unknown said...

UK42 here, i just found my two halves - One is among the northern French i believe, a bit south of the main British Isles cluster, and the other half is close to a blob just east of the main British Isles cluster. Although i am only slightly mixed (with a North African or Sephardic Jewish population), i would have expected one half to be somewhere in the main British cluster and the other half to be somewhere to the south-east of that, and split along a NW-SE axis, although it seems to be split along a SW-NE axis instead.

Do you think that this just happens to be down the fact that it can come up with different solutions for the same person as you say? Otherwise it is a little confusing.

Sam Jackson

Michael said...

Hi David,

GRDE1 here. I am surprised by how spread out the two halves were for mine. One appears to the far west with Sephardic Jews (my mother was Ashkenazic) and the other with Balkars (My father was Greek and German (from Poland)). How should I interpret the data?

Thank you.

Cantillon said...

Hi david,

Sorry but one last time please. I believe I'm UK43, I was Au5. Is this correct? thanks

Guddarhagen J├╝rgen said...

hello Admin,
how can I place myself on this map?

Davidski said...

You can't, but you can try this stuff...