How does autosomalDNA calculators calculate what ethnicity you belong to?

[Konpirike]

How does autosomalDNA calculators calculate what ethnicity you belong to?

Let me give you an example about what I mean through admixture percentage:

Let's say I am british and there is a K3 calculator that has 3 components. One is North European,one is South European and one is East Asian.

Let's say a typical british person gets 80%North European,20%South European and 0%East Asian on this k3 calculator. But eventhough I am British person I get 85%North European,10%South European and 5% East Asian which looks more closer to the average Finnish person.

Will the calculator in this way try to figure out what ethnicity I belong to(through admixture percentage), or will the calculator still make me come out as British, then I suppose , if the latter,then it uses some kind of a relative matching tool which in this way figures out your ethnicity I guess.


So is it through admixture percentage or is it through calculating what ethnicity most of your "relatives" belong to?

Anybody that knows?

[Sokoli]

How does autosomalDNA calculators calculate what ethnicity you belong to?

Let me give you an example about what I mean through admixture percentage:

Let's say I am british and there is a K3 calculator that has 3 components. One is North European,one is South European and one is East Asian.

Let's say a typical british person gets 80%North European,20%South European and 0%East Asian on this k3 calculator. But eventhough I am British person I get 85%North European,10%South European and 5% East Asian which looks more closer to the average Finnish person.

Will the calculator in this way try to figure out what ethnicity I belong to(through admixture percentage), or will the calculator still make me come out as British, then I suppose , if the latter,then it uses some kind of a relative matching tool which in this way figures out your ethnicity I guess.


So is it through admixture percentage or is it through calculating what ethnicity most of your "relatives" belong to?

Anybody that knows?

Genes have no idea about ethnicity. Genetic tests measure how genetically close ore distant are any two individuals. Because marriages are more frequent within an ethnicity than among different enthicities, people that share ethnicity will share genes as well and will plot close to each other. Your ethnicity is the ethnicity reported by other people that have taken the test genetically close to you. The same way is calculated how close you're with other North / South Europeans, Asians, etc.

[Konpirike]

Genes have no idea about ethnicity. Genetic tests measure how genetically close ore distant are any two individuals. Because marriages are more frequent within an ethnicity than among different enthicities, people that share ethnicity will share genes as well and will plot close to each other. Your ethnicity is the ethnicity reported by other people that have taken the test genetically close to you. The same way is calculated how close you're with other North / South Europeans, Asians, etc.

Hmmm, that's also what I thought was the most logical. But then here is why I got doubts about it. For example on GEDmatch I have seen people that belong to one ethnicity, but get grouped with another ethnicity just because they don't match with the average scoring of that ethnicity which they actually belong to. But then when we look at that persons relatives, most of his relatives belong to the given ethnicity which he claims. That is why I find it confusing. I have seen it more then a couple of times. It is abit strange.

[Dr. Robotnik the Subbotnik]

Excellent question. I also would like to know about this.

On some calculators my top pop. is Hungarian (I think this makes sense if they are basing it purely off WHG/ENF/etc. levels - as I have as much ENF/WHG as a typical Hungarian person, probably from my stupid fucking Rhenish ancestors that gave me too much woggy ENF - plus Hungarians have a little dash of East-Asian like I do, it makes them the perfect fit for me).

However, to my knowledge I haven't a single drop of Hungarian in me, so, this calc. must have been basing it off of percentages of WHG-ENF-East Asian, etc.

On other calculators, I get North-German as my closest pop. which I don't really understand.

[Konpirike]

Excellent question. I also would like to know about this.

On some calculators my top pop. is Hungarian (I think this makes sense if they are basing it purely off WHG/ENF/etc. levels - as I have as much ENF/WHG as a typical Hungarian person, probably from my stupid fucking Rhenish ancestors that gave me too much woggy ENF - plus Hungarians have a little dash of East-Asian like I do, it makes them the perfect fit for me).

However, to my knowledge I haven't a single drop of Hungarian in me, so, this calc. must have been basing it off of percentages of WHG-ENF-East Asian, etc.

On other calculators, I get North-German as my closest pop. which I don't really understand.

Exactly you are a very good example!

I have seen people that are 50/50. For an example 50% Greek and 50%Swedish but get German as the the closest pop.

[Damião de Góis]

They just compare your results to their set of samples. The key is to look at several calculators and pay attention to distances.

I would say that if you have the same consistently the same population at a distance lower than 5 on several runs such as Eurogenes EUTest, Eurogenes K13 or Dodecad V3 then there's a big chance you're from that ethnicity. At least it works for me.

[Antimage]

Excellent question. I also would like to know about this.

On some calculators my top pop. is Hungarian (I think this makes sense if they are basing it purely off WHG/ENF/etc. levels - as I have as much ENF/WHG as a typical Hungarian person, probably from my stupid fucking Rhenish ancestors that gave me too much woggy ENF - plus Hungarians have a little dash of East-Asian like I do, it makes them the perfect fit for me).

However, to my knowledge I haven't a single drop of Hungarian in me, so, this calc. must have been basing it off of percentages of WHG-ENF-East Asian, etc.

On other calculators, I get North-German as my closest pop. which I don't really understand.

With the nordic ancestry you have, how can you have same ENF/WHG as a hungarian person?

[Dr. Robotnik the Subbotnik]

They just compare your results to their set of samples. The key is to look at several calculators and pay attention to distances.

I would say that if you have the same consistently the same population at a distance lower than 5 on several runs such as Eurogenes EUTest, Eurogenes K13 or Dodecad V3 then there's a big change you're from that ethnicity. At least it works for me.

Using your approach, I am either Orcadian (top pop. on Eurogenes EUTest, distance is 3.27), North German (top pop. on Eurogenes K13, distance is 4.16) or CEU (HapMap) (top pop. on Dodecad V3, distance is 2.97).

I don't know what CEU (HapMap) is lol

[Dr. Robotnik the Subbotnik]

With the nordic ancestry you have, how can you have same ENF/WHG as a hungarian person?

As I said because of my Rhenish ancestry. I am not fully Nordic. I am mixed Irish, Nordic, Rhenish, maybe Austrian, English..

[Antimage]

As I said because of my Rhenish ancestry. I am not fully Nordic. I am mixed Irish, Nordic, Rhenish, maybe Austrian, English..

Still all of those have more whg than hungarians.

[Damião de Góis]

Using your approach, I am either Orcadian (top pop. on Eurogenes EUTest, distance is 3.27), North German (top pop. on Eurogenes K13, distance is 4.16) or CEU (HapMap) (top pop. on Dodecad V3, distance is 2.97).

I don't know what CEU (HapMap) is lol

Those distances are low enough to tell you're very close to those ethnicities. They also cluster very near each other, so you'll get different populations.

CEU is Utah Germans i think.

[Dr. Robotnik the Subbotnik]

Still all of those have more whg than hungarians.

Not true, a Rhenish person from more by France will have same WHG/ENF levels as Hungarians, just minus the minor East-Asian from Magyars. It is said Hungarians score similarly to South-Germans (which does include southern Rhineland) minus the bit of East-Asian. I know what I am talking about.

[Dr. Robotnik the Subbotnik]

Those distances are low enough to tell you're very close to those ethnicities. They also cluster very near each other, so you'll get different populations.

CEU is Utah Germans i think.

I wonder from which part of Germany Utah Germans predominantly hail.

Also on some calcs I get Hungarian as my top pop.

as far as my ENF/WHG levels, I have higher ENF than even Brits (on level with Hungarians, Southern Germans). I have a bit of East-Asian though (2%) as well.

[Dick]

Not true, a Rhenish person from more by France will have same WHG/ENF levels as Hungarians, just minus the minor East-Asian from Magyars. It is said Hungarians score similarly to South-Germans (which does include southern Rhineland) minus the bit of East-Asian. I know what I am talking about.
+1

[Dr. Robotnik the Subbotnik]

In conclusion I think as far as genetic relationship I am closest to Orcadians, North Germans, Utah Germans, but as far as percentages of the ancient populations (WHG, ENF, East-Asian (Uralic)) I would have about the same percentages as a Hungarian, this is probably the only way to explain my results.

[Sokoli]

Exactly you are a very good example!

I have seen people that are 50/50. For an example 50% Greek and 50%Swedish but get German as the the closest pop.

Makes sense. If you make an imaginary line between Greek and Swede samples, someone that is 50/50 will plot in the middle of that line. Which happens to be where Austrians and some Germans plot.
http://blogs.discovermagazine.com/gnxp/files/2011/05/sobpc.jpg

[de Burgh II]

Your genes merely manifest through mathematical approximations as a means to break down one's autosomal dna via their reference samples/population.

If an individual is mixed in terms of ancestry, then it will find a median/middle proxy in their reference samples.

For example, say an individual is roughly 50% German and 50% Northern Italian; the calculator will try to find logistical proxy to estimate where you fall within the middle of these two places of origin. So, if we go by this simple heuristic for this example, the individual should pop out as being Austrian or Swiss logically speaking as their middle proxy.

[Voskos]

it calculates your iq and places you on a PCA plot.

[firemonkey]

Using EU test ,Eurogenes K13 and Dodecad v3 I am Orcadian(2.299) West Scottish(2.074) CEU(2.142) using FTDNA data. I get Hungarian as first population on a few calculators. I think quite a few people with British/Irish ancestry, like me, have reported this.

[Neon Knight]

I think CEU is Utah whites in general, but they are mostly from British stock.

My closest Oracle result is on Dodecad V3 where I get 0.88 Argyll (West Scotland). But I get 10.68 for Kent (SE England) and they really should not be so far apart.

So why the inconsistency? It looks to me like it is not sampling enough genetic variation to be accurate.

[firemonkey]

I think CEU is Utah whites in general, but they are mostly from British stock.

My closest Oracle result is on Dodecad V3 where I get 0.88 Argyll (West Scotland). But I get 10.68 for Kent (SE England) and they really should not be so far apart.

So why the inconsistency? It looks to me like it is not sampling enough genetic variation to be accurate.

I see your point with Dodecad V3 Oracle 4 least squares method I get: 1 CEU @ 2.141554
2 Orcadian @ 2.491170
3 Argyll @ 2.645659
4 Orkney @ 2.665827

11 Kent @ 12.793219
12 British_Isles @ 14.495218
13 British @ 14.792280
14 Cornwall @ 15.587780

Doesn't seem right to me.

[decordoba]

I am Austrian - my 23andMe result is - 78 % Northwestern European, 9,5 % Southern European, 0,5 % Eastern European and 12 % Broadly European.

My Eurogenes K13 Oracle calculation is different

NorthAtlantic ........... 42 %
Baltic ...................... 21 %
West_Med ................ 18 %
East_Med ................ 11 %
West_Asian ............... 6 %
South_Asian .............0,6 %

------------------------------------------

# Population (source) Distance
1 French ............. 4.28
2 West_German .... 4.6
3 South_Dutch ...... 5.08
4 Southeast_English 10.14
5 Austrian ............. 10.27

[Damião de Góis]

Dodecad V3 had fewer populations.
Anyway i get:

Eurogenes K13:
1 Portuguese 2.47

Eurogenes EUTest:
1 PT 3.63

Dodecad V3:
1 Portuguese (Dodecad) 3.69

I consider Dodecad k12b and Eurogenes K15 to be good runs too, but in those i get:

K12b:
1 Extremadura (1000Genomes) 5.89

K15:
1 Spanish_Castilla_Y_Leon 3.13

[firemonkey]

K12b - 1 Dutch @ 1.59

K15 - 1 Irish @ 2.17

[PunhetaDeBacalhau]

I think CEU is Utah whites in general, but they are mostly from British stock.

My closest Oracle result is on Dodecad V3 where I get 0.88 Argyll (West Scotland). But I get 10.68 for Kent (SE England) and they really should not be so far apart.

So why the inconsistency? It looks to me like it is not sampling enough genetic variation to be accurate.

Basically, calculators that bunch different components together will never have very accurate oracle results.

Dodecad V3's West European is basically West European + North European, meaning it doesn't distinguish between the two. Same thing with Eurogenes K13, North Atlantic is west european + north european, and the oracle results tend to not be very accurate, as a person that is 40% western + 5% northern will score the same North Atlantic as a 5% western + 40% northern person. Same thing goes for all calculators that bunch components up together that shouldn't be bunched up.

I'm sure Eurogenes K15, that distinguishes between west and north european, gives you better oracle results, no? EUTest distinguishes it too. But the best I've used for oracle results is MDLP k23b, not because the component names and percentages are very correct, but because it has so many components that the oracle works really well at distinguishing between ethnicities (i suppose even between northern europeans, which is generally very hard). Generally, more components = more oracle accuracy.

It's a shame that Eurogenes K36 doesn't have an oracle. If it did, it would have by far the best oracle results of any calculator. Dunno why it doesn't. Maybe because calculating oracle results using 36 components takes quite a bit of time and processing power (you will notice that the oracle for MDLP k23b also takes quite a bit more time to calculate than other calculators with less components). But the results would certainly be very accurate.

[decordoba]

Basically, calculators that bunch different components together will never have very accurate oracle results.


I'm sure Eurogenes K15, that distinguishes between west and north european, gives you better oracle results, no? EUTest distinguishes it too. But the best I've used for oracle results is MDLP k23b, not because the component names and percentages are very correct, but because it has so many components that the oracle works really well at distinguishing between ethnicities (i suppose even between northern europeans, which is generally very hard). Generally, more components = more oracle accuracy.

It's a shame that Eurogenes K36 doesn't have an oracle. If it did, it would have by far the best oracle results of any calculator. Dunno why it doesn't. Maybe because calculating oracle results using 36 components takes quite a bit of time and processing power (you will notice that the oracle for MDLP k23b also takes quite a bit more time to calculate than other calculators with less components). But the results would certainly be very accurate.

Thank you for the good hints !

My MDLP k23 result:

# Population Percent
1 European_Hunters_Gatherers ... 33.7
2 European_Early_Farmers ...... 29.53
3 Caucasian .......................... 27.39
4 Ancestral_Altaic ................. 3.52
5 South_Central_Asian ............ 3.48
6 North_African .................... 1.33
7 Archaic_Human ................... 0.44
8 Tungus-Altaic .................... 0.31
9 Australoid .......................... 0.3

On my mind - this calculator shows the rest of IE-ancestry (Altaic, South_Central Asian,..), the other parts of the IE-admixture are included in WHG, CHG, Neolithic_Farmers.

Thank you to the creator and DNA-expert for this calculator !

[decordoba]

Add to my MDLP k23 result:

Genetic distance:

# Population (source) Distance
1 Italian_North ( ) ............. 5.94
2 German-Volga ( ) ................7.19
3 South_German ( )........ 7.31
4 Frisian ( ) ................... 9.65
5 Belgian ( ).................. 9.77
6 Austrian ( )................ 10.09
7 Italian_Bergamo ( )....... 10.15
8 Dutch ( ) ................... 10.99
9 Italian_Piedmont ( )...... 11.06
10 North_German ( ) ......... 11.48
11 Irish ( ) ..................... 11.9

[PunhetaDeBacalhau]

Add to my MDLP k23 result:

Genetic distance:

# Population (source) Distance
1 Italian_North ( ) ............. 5.94
2 German-Volga ( ) ................7.19
3 South_German ( )........ 7.31
4 Frisian ( ) ................... 9.65
5 Belgian ( ).................. 9.77
6 Austrian ( )................ 10.09
7 Italian_Bergamo ( )....... 10.15
8 Dutch ( ) ................... 10.99
9 Italian_Piedmont ( )...... 11.06
10 North_German ( ) ......... 11.48
11 Irish ( ) ..................... 11.9

Are you fully Austrian? Maybe the Austrian average on that Oracle is made of pseudo-Austrian samples. Anyway, the first results you get are very close to home...

To me, It's the only calculator that successfully separates Portugal from Spanish Galicia and Spanish Extremadura. Almost all the others put me as a Galician.

# Population (source) Distance
1 Portugese ( ) 2.57
2 Spaniard ( ) 6.98
3 Spanish_Extremadura_IBS ( ) 7.17
4 French ( ) 7.46
5 Spanish_Galicia_IBS ( ) 7.94
6 Spanish_Baleares_IBS ( ) 8.24
7 Spanish_Murcia_IBS ( ) 8.51
8 Spanish_Castilla_y_Leon_IBS ( ) 8.68
9 Spanish_Cataluna_IBS ( ) 9.14
10 Spanish_Canarias_IBS ( ) 10.05

[Neon Knight]

Basically, calculators that bunch different components together will never have very accurate oracle results.

Dodecad V3's West European is basically West European + North European, meaning it doesn't distinguish between the two. Same thing with Eurogenes K13, North Atlantic is west european + north european, and the oracle results tend to not be very accurate, as a person that is 40% western + 5% northern will score the same North Atlantic as a 5% western + 40% northern person. Same thing goes for all calculators that bunch components up together that shouldn't be bunched up.

I'm sure Eurogenes K15, that distinguishes between west and north european, gives you better oracle results, no? EUTest distinguishes it too. But the best I've used for oracle results is MDLP k23b, not because the component names and percentages are very correct, but because it has so many components that the oracle works really well at distinguishing between ethnicities (i suppose even between northern europeans, which is generally very hard). Generally, more components = more oracle accuracy.K13, K15 & EU Test all give me very similar results, not really distinguishing between British, Irish, Dutch, German & Danish. MDLP K23b gives me closer values but is still a bit mixed up with the countries:

01 Frisian @ 1.670166
02 Irish @ 2.176106
03 English @ 2.219811
04 Belgian @ 2.308079
05 North_European @ 3.103471
06 Dutch @ 3.556246
07 English_Kent_GBR @ 4.482354
08 CEU @ 4.486952
09 British @ 5.067211
10 English_Cornwall_GBR @ 5.146961
11 South_German @ 5.196607
12 Welsh @ 5.652264
13 German-Volga @ 5.875906
14 Scottish_Argyll_Bute_GBR @ 5.907039

Do you think ancestry testing will improve in future?

[Damião de Góis]

I have a problem with MDLP. First i can't get over the fact that they can't spell "portuguese" or "provençal", and finally it guesses me all over the place on different runs.

MDLP World:


# Population (source) Distance
1 Swiss 5.05
2 Provancal 5.33
3 Italian_North 5.39
4 Portugese 6.33
5 Montenegrin 7.01
6 Macedonian 7.82
7 Bulgarian 7.85
8 Serbian 8.84
9 Italian-North 8.87
10 Iberian 8.95

MDLP World 22:


# Population (source) Distance
1 Swiss (derived) 4.51
2 Italian_North (derived) 4.85
3 Provancestralal (derived) 5.6
4 Portugese (derived) 8.45
5 French (derived) 9.41
6 Montenegrin (derived) 9.51
7 Italian-North (derived) 9.76
8 Iberian (derived) 9.93
9 Bulgarian (derived) 10.05
10 Romania (derived) 10.12

MDLP k23b:


# Population (source) Distance
1 Belgian ( ) 8.9
2 South_German ( ) 9.05
3 English ( ) 9.67
4 Frisian ( ) 9.68
5 French ( ) 9.7
6 English_Kent_GBR ( ) 9.72
7 German-Volga ( ) 10.06
8 Irish ( ) 10.37
9 Portugese ( ) 10.46
10 North_European ( ) 10.61

[PunhetaDeBacalhau]

K13, K15 & EU Test all give me very similar results, not really distinguishing between British, Irish, Dutch, German & Danish. MDLP K23b gives me closer values but is still a bit mixed up with the countries:

01 Frisian @ 1.670166
02 Irish @ 2.176106
03 English @ 2.219811
04 Belgian @ 2.308079
05 North_European @ 3.103471
06 Dutch @ 3.556246
07 English_Kent_GBR @ 4.482354
08 CEU @ 4.486952
09 British @ 5.067211
10 English_Cornwall_GBR @ 5.146961
11 South_German @ 5.196607
12 Welsh @ 5.652264
13 German-Volga @ 5.875906
14 Scottish_Argyll_Bute_GBR @ 5.907039

Do you think ancestry testing will improve in future?

Yea, the components of the calculator seem to be not as good to distinguish Northern Europeans as they are for Southern Europeans. Also, seems the samples being used for Oracle comparisons in the MDLP project don't represent very well the populations they are supposed to represent.

Surely it will improve, but unfortunately the only gedmatch calculators released recently have been based around ancient times populations. I think we need a calculator based only on providing the best oracle results possible, regardless of component percentages it gives. K36 shows me as being 17% Northern Italian&Tuscan, which i'm not, but it makes sense since the kind of ENF that they have is very similar to Iberian ENF. K36 components should be seen more as "connections" to other populations rather than actual recent ancestry from those populations, and IMO would give great Oracle results. When I have more time, I'll try to make a calculator similar to K36 that includes an Oracle.

[PunhetaDeBacalhau]

I have a problem with MDLP. First i can't get over the fact that they can't spell "portuguese" or "provençal", and finally it guesses me all over the place on different runs.

MDLP World:



MDLP World 22:



MDLP k23b:

Wow, those are some very bad oracle results... And yes, we can't really trust results of calculators made by people that can't spell "PortugUese", I agree :P I expected bad results from both the MDLP Worlds (aggregating components like Southern European and Western European into a single component called "South_and_West European" could only go wrong), but not so much the k23b... I get Portuguese first in all three of them, but both World's give me Portuguese at a distance of 8, probably due to bad Portuguese samples.

[Damião de Góis]

Wow, those are some very bad oracle results... And yes, we can't really trust results of calculators made by people that can't spell "PortugUese", I agree :P I expected bad results from both the MDLP Worlds (aggregating components like Southern European and Western European into a single component called "South_and_West European" could only go wrong), but not so much the k23b... I get Portuguese first in all three of them, but both World's give me Portuguese at a distance of 8, probably due to bad Portuguese samples.

I find it strange that you are happier with MDLP k23b than with any of the Eurogenes or Dodecad runs. I was a part of Eurogenes and Dodecad original PT samples, but not of MDLP, don't know if this is important for Oracle or not.

[PunhetaDeBacalhau]

I find it strange that you are happier with MDLP k23b than with any of the Eurogenes or Dodecad runs. I was a part of Eurogenes and Dodecad original PT samples, but not of MDLP, don't know if this is important for Oracle or not.

I didn't say I was happier with the MDLP calculators overall, as IMO they are kinda bad. Eurogenes and Dodecad give you better percentages of components that you actually have, but the components are too broad to distinguish between close populations in the Oracle. What i did say is that the Oracle of the MDLP k23b gave me the best result by far of all the calculators (distinguished very well between Portuguese and Spaniard), and that's due to the higher number of components the calculator has in comparison to other ones.

[Damião de Góis]

I didn't say I was happier with the MDLP calculators overall, as IMO they are kinda bad. Eurogenes and Dodecad give you better percentages of components that you actually have, but the components are too broad to distinguish between close populations in the Oracle. What i did say is that the Oracle of the MDLP k23b gave me the best result by far of all the calculators (distinguished very well between Portuguese and Spaniard), and that's due to the higher number of components the calculator has in comparison to other ones.

Ok, just out of curiosity, do you get similar Oracle results to me on these Eurogenes and Dodecad runs?

http://www.theapricity.com/forum/showthread.php?190530-How-does-autosomalDNA-calculators-calculate-what-ethnicity-you-belong-to&p=3966581&viewfull=1#post3966581

I don't consider a guess of Castilla Y Leon, Extremadura or Galicia to be that bad, because we are basically being guessed as western iberians which is still accurate. As long as the distances are below 5 it's ok.

[PunhetaDeBacalhau]

Ok, just out of curiosity, do you get similar Oracle results to me on these Eurogenes and Dodecad runs?

http://www.theapricity.com/forum/showthread.php?190530-How-does-autosomalDNA-calculators-calculate-what-ethnicity-you-belong-to&p=3966581&viewfull=1#post3966581

I don't consider a guess of Castilla Y Leon, Extremadura or Galicia to be that bad, because we are basically being guessed as western iberians which is still accurate. As long as the distances are below 5 it's ok.

I get:



Eurogenes K13 (puts Atlantic and North Sea into the same bag, so the fact that we have slightly higher North Sea due to the Suebi and less Atlantic due to being further away from Basques gets ignored):

1 Spanish_Murcia 4.77
2 Spanish_Castilla_Y_Leon 4.8
3 Spanish_Extremadura 5.09
4 Spanish_Castilla_La_Mancha 5.37
5 Spanish_Galicia 5.58
6 Portuguese 5.63

Eurogenes K15 (Does a better job than K13 for me. IMO it takes too much from West Med and puts it into Atlantic, and the North Sea is too high. The reason I start with 6 distance in this one is because the calculator makes a mistake and assigns the normal 4% Eastern European that the Portuguese get into North Sea, making me a ridiculous 26.26 North Sea and 0% East Euro):

1 Spanish_Galicia 6.14
2 Portuguese 6.52
3 Spanish_Extremadura 7.01
4 Spanish_Castilla_Y_Leon 7.44
5 Spanish_Murcia 7.54

EUTest (My favorite Eurogenes calculator, doesn't seem to make any silly error. Shame it doesn't distinguish Spanish regions to see if i would get PT first anyway):

# Population (source) Distance
1 PT 4.08
2 ES 5.4
3 FR 10.18
4 North_Italian 11.95
5 Tuscan 14.49

Dodecad V3 (very bad, gives 10 distance to Portuguese, and thinks I'm closer to France than Spain):

1 Portuguese (Dodecad) 10.12
2 French (Dodecad) 12.74
3 French (HGDP) 13.16
4 IBS (1000Genomes) 13.8
5 N_Italian (Dodecad) 13.93
6 Spaniards (Behar) 14.03

Dodecad k12b:

1 Galicia (1000Genomes) 6.19
2 Portuguese (Dodecad) 6.28
3 Extremadura (1000Genomes) 6.81
4 Baleares (1000Genomes) 8.07
5 Castilla_Y_Leon (1000Genomes) 9.24



Funny thing is that i get Galicia a lot and I'm not even Northern Portuguese, I'm from the center :blink:

Anyway, it can't distinguish well between Western Iberians, which are very close populations. Same thing happens to Northern Europeans. But I believe a K36 Oracle would be able to.

[decordoba]

[QUOTE=PunhetaDeBacalhau;3969845]Are you fully Austrian? Maybe the Austrian average on that Oracle is made of pseudo-Austrian samples. Anyway, the first results you get are very close to home...
To me...

I am Austrian, but I was born at the border to Bavaria/South Germany. This area was part of Bavaria 250 years ago. And one Grandmother - her family came from the Alps long time ago (maybe North_Italy today?).

The Austrian average: maybe includes a high amount of Slavic people, because the eastern part of Austria is partly Slavic area.