Sunday, February 24, 2013

Let's play with Oracle-4

For a brief remainder :

- 'K7b' results : "Dienekes' K7b calculator : my results"
Both a remote genetic cousin of mine from West Béarn and myself from East Béarn do differ from mainstream French for low West Asian admixture which peaks in the Caucasus.

- 'Oracle' results : "My Oracle results (Population Sharing)"
The algorithm finds - for two different calculators - that I'm closest to Spanish Basques, then to French Basques, then to Cantabrians and Aragonese people.

- 'globe10' results : "Dienekes' globe10 calculator : my results"
Basically speaking, I can be analyzed as the succession of 3 layers : 76% Atlantic-Baltic (peaking in Mesolithic Asturians) / 21% Southern (peaking in Palestinians, probably linked with Neolithic migrations) / 2% West Asian (peaking in the Caucasus, Indo-Europeans ?).

- "D-statistics on ADMIXTURE components"
Are my minor components from ADMIXTURE runs real or just algorithmic artefacts ?

Now, let's deal with Oracle-4 : the philosophy behind that tool is the same that behind Oracle except results are more complete. Let's see them when used with Dienekes' world9 calculator :

I . Admix Results :

1 Atlantic_Baltic 73.55
2 Southern 23.88
3 Caucasus_Gedrosia 2.57

The three main "layers" are to be found again, with different values according to calculators and samples used : world9 results are pretty to similar to world10 results. I also show 0.88% Amerindian and 0.63% South Asian but they are probably artefacts.

Let's compare these results with populations of interest for me : see Dienekes' world9 spreadsheet for other results.

1. Spanish Basques :


2. French Basques :


Both Basque populations are virtually similar. Very low Caucasian admixture. Both populations also show minor Amerindian and South Asian but those are probably artefacts.

I'm thus confirmed to be higher on Caucasian admixture than Basque people (circa 2,6% against less than 1% for Basque people). I believe this admixture is real and is not an artefact : my home area was more opened to the latest waves of migrants coming from the East, even though just marginally.

3. French people :

What does "French" mean ? Nothing. That free-access sample has been used for years on the Internet : allegedly it was captured in Lyon (SE France, Rhône valley) but when analyzed (for instance the famous 23andme graph), there are some disparities that hint to the fact those French people are of varied origins (probably a mix of Rhodanian and Alpine people with extra Provençal and Burgundian imput : this is the basis of the peopling of modern Lyon).

Nevertheless, about 10% of Caucasian admixture is high in regards to my own results : on average, the French, whatever that means, have been strongly affected by Bronze Age migrants and Indo-European migrations. Add that those migrants reaching Western Europe were already mixed through their journey from East to West and consequently already showed strong numbers in the main two components. The impact of such migrations should then be higher than the sole Caucasian admixture result.

Not taking Caucasian admixture into account, the Atlantic/Southern ratio is around 3.7 for the French. It is around 3 for my own results just like the Basques. That means that French people from Lyon were less affected by previous southern migrations (Neolithic ?) which might have brough agriculture than the Basques and myself.

4. Aragonese people :


What does "Aragonese" mean ? Nothing : Aragon is pretty varied and one should expect people from Jacetania in the Pyrenees to differ much from people in the Ebro valley or from Teruel.

Still, the results are interesting : Aragonese people show a relatively low Caucasian admixture within the European context which is coherent with what we know of the pre-Indo-European character of this land (either Vasconic or Iberian).

The Atlantic/Southern ratio is 2.4 : the area from where those people Aragonese people were sampled were more strongly affected by Neolithic migrations than Basque people and the French. I suppose one can guess the sample is from around the Ebro valley.

5. Cantabrians :


Cantbrians show results rather similar to Aragonese people : 5% of Caucasian still is rather low when compared with the French but is very high when compared to neighbouring Basque people.We know Cantabria was indo-europeanized as proved by ancient onomastics : this is one genetic proof.

The Atlantic/Southern ratio is 2.5, on par with Aragonese results. Let's notice that Cantabria shows non-negligible African admixture (around 0.8%), probably a proxy for North African admixture that I dare link with the many forth and back migrations between Iberia and North Africa in prehistoric times.

The reasons why Basque lands escaped such migrations remain a mystery : Cantabria and the Basque Country are pretty similar geographically wise. More local samples would be needed : people from Las Encartaciones in Biscay might differ from mainstream Basque people for instance. Conversely, Pas people in Cantabria might differ from people in Campoo.

6. Catalan people :


Unsurprisingly enough, Catalan people appear to be more affected by Caucasian admixture than fellow Cantabrian and Aragonese people : Catalonia is more opened to foreign influence. The Atlantic/Southern ratio is also 2.5, on par with Aragon.

If possible, more local samples would be needed : people from Pyrenean Catalonia clearly must differ from people in Tarragona.

7. Other Iberians :

You can browse the spreadsheet for other Iberians' results. Main trends are the following ones :

- As one goes southwards, the Atlantic/Southern ratio is lower : southern areas of Iberia were more affected by Neolithic migrations than northern areas.

- Some areas do show substantial African admixture as defined by this algorithm (part of true North African admixture might be hidden since an African component is used as a proxy by the algorithm) : 2,1% for Extremadura, 1,6% for Galicia, 4,1% for Canary Islands.

II . Population approximations :

Here are my various results.

1. Using 1 population approximation :

1 Pais_Vasco @ 1.983
2 French_Basque @ 2.307
3 Cantabria @ 7.640
4 Aragon @ 8.369
5 French @ 9.158
6 Cataluna @ 9.231
7 Spanish @ 10.120
8 Valencia @ 10.268
9 French @ 10.682
10 Spaniards @ 11.639

The results are not surprising, they have been quite consistent since such tools exist : I'm closest to Basque people, something which could be inferred by my admixture results. My only difference with Basque people seem to stem from a higher West Asian admixture which seems pretty logical considering my area at the feet of the Pyrenees never was as isolated as deep Basque valleys.

Then come Cantabrians and Aragonese people aka peripheral Vasconic people : despite the undeniable Celtic past of Cantabria and what appears to be a sharp genetic rift somewhere near the border with Biscay, Cantabrian people still are close to other Vasconic people.

One can theoricize that Asturian people would be somewhere inbetween Cantabrians and Galicians. Nevertheless, the Iberian peninsula is confirmed to be divided into two parts by a line which would go from Asturias to Murcia.

2. Using 2 populations approximation :

1 50% Pais_Vasco +50% Pais_Vasco @ 1.983
2 50% French_Basque +50% Pais_Vasco @ 2.140
3 50% French_Basque +50% French_Basque @ 2.307
4 50% French_Basque +50% Cantabria @ 3.391
5 50% Pais_Vasco +50% Cantabria @ 3.467
6 50% French +50% French_Basque @ 3.469
7 50% French +50% Pais_Vasco @ 3.667
8 50% French_Basque +50% Aragon @ 3.827
9 50% French_Basque +50% Cataluna @ 3.874
10 50% Aragon +50% Pais_Vasco @ 3.892

I'm best explained as being the result of two Spanish Basques. The results are not that interesting : the algorithm seems to try coping with my West Asian admixture, hence why it uses Cantabrians then French as a way to balance the low West Asian imput in the Basques.

3. Using 4 populations approximation :

1 French + French_Basque + French_Basque + French_Basque @ 0.734
2 French + French_Basque + French_Basque + Pais_Vasco @ 0.833
3 French + French_Basque + Pais_Vasco + Pais_Vasco @ 0.934
4 French + Pais_Vasco + Pais_Vasco + Pais_Vasco @ 1.037
5 French + French_Basque + French_Basque + French_Basque @ 1.081
6 French + French_Basque + French_Basque + Pais_Vasco @ 1.179
7 French + French_Basque + Pais_Vasco + Pais_Vasco @ 1.280
8 French + Pais_Vasco + Pais_Vasco + Pais_Vasco @ 1.382
9 French_Basque + French_Basque + French_Basque + Cataluna @ 1.613
10 French_Basque + French_Basque + Cataluna + Pais_Vasco @ 1.634
11 French_Basque + Cataluna + Pais_Vasco + Pais_Vasco @ 1.663
12 Cataluna + Pais_Vasco + Pais_Vasco + Pais_Vasco @ 1.699
13 Lithuanian + French_Basque + French_Basque + Sardinian @ 1.745
14 French_Basque + French_Basque + Sardinian + Lithuanians @ 1.750
15 Pais_Vasco + Pais_Vasco + Pais_Vasco + Cantabria @ 1.787
16 French_Basque + Pais_Vasco + Pais_Vasco + Cantabria @ 1.792
17 French_Basque + French_Basque + Pais_Vasco + Cantabria @ 1.804
18 French_Basque + French_Basque + French_Basque + Cantabria @ 1.823
19 Lithuanian + French_Basque + Sardinian + Pais_Vasco @ 1.839
20 French_Basque + Sardinian + Lithuanians + Pais_Vasco @ 1.845

The first approximation seems to be quite precise : to sum up, my genetic background is one of a Basque individual switched towards France or Catalonia. Nothing surrealistic here. Quite boring actually ! People who know their background cannot expect surprises.

Nevertheless, when having a look at these results, I can only wish France one day will get interested in genetic studies and that we will be able to access new samples.

D-statistics on ADMIXTURE components

Aka I'm the subject of an article by Dienekes ! : )
Well, DOD133 actually.

I don't get all the subtleties, but as far as I understand the article, it's about detecting whether minor components from ADMIXTURE runs are real or just algorithmic artefacts :

One of the most persistent questions I get as admin of the Dodecad Project is whether some low level of admixture (e.g., 0.7%) of some ancestral component is "noise" or "real".

You can check the article here :


Thursday, February 21, 2013

Carnia : Friuli (Italy)

Carnia is located south of the main chain of the Carnic Alps, in the northwest of the Udine province.

  •  Sample :

Full size

  • Analyzis :

- Brachy type : the following individuals are part of a classical North Italian variation, many of them do exhibit blondism and sharp features.

Another series :

- Lepto type : pan-North Italian (sharp features, light features).

  • Morphotypes :

Tuesday, February 19, 2013

La Gomera : Canary Islands (Spain)

According to a 2011 study, autochtonous Guanche autosomal ancestry is at its highest in La Gomera, circa 40%.

  • Sample :

Full size

  • Brief anthropological analyzis :

- Brachy type : some individuals may show Berber features but all in all those people look Andalusian (wide-spaced sloping and rather chinky eyes, short and sometimes broad nose, well-defined mouth area, ...)

- Lepto type : once more some individuals may show Berber features but all in all those people look Andalusian within a pan-Iberian variation (narrow "horsy" face, long and straight nose yet somehow large, a rather broad jaw, close-set eyes, ...)

  • Final morphotypes :

Tuesday, February 5, 2013

Aracena : Andalusia (Spain)

  • Sample :

Full size

  • Brief anthropological analyzis :

- Brachy type : striking Andalusian features are a puffy nose and a well-defined mouth area ;



- Lepto type : pan-Iberian ;

  • Morphotypes :