Geographic Population Structure (GPS) prediction

[MfA_]

New tool to play around :)


Actually, to run GPS users can input what seem to be Geno 2.0 autosomal ancestry proportions into the relevant fields on this page (http://mendel.fat.glam.ac.uk/projects/gpsNGG2/personal.php). Now, for various reasons I don't have a very high opinion of the Geno 2.0 autosomal test, so it's not something I'll ever pay for. However, I just quickly put together a K=9 test that roughly approximates the Geno 2.0 analysis. It's a bit noisy, but here's the GPS result it generated (using 10 reference populations), which is fairly accurate...

http://img593.imageshack.us/img593/2489/oe0e.png

The files for the K=9 test are available here (https://docs.google.com/file/d/0B9o3EYTdM8lQVmZlcEYyTFFzeHM/edit?usp=sharing), but I can't guarantee they'll produce accurate results for everyone. My outcome might have been a bit of a fluke. I'd say the best thing to do is to wait until the site starts accepting raw genotype data from users (see here).


http://polishgenes.blogspot.com/2013/09/coming-soon-geographic-population.html

Download K9 files, create a text file copy the followings and save as k9.par..


1d-7
9
genotype.txt
121122
K9_clusters.txt
K9_frequencies.txt
K9_alleles.txt
verbose
genomewide

Edit: k9.par (https://www.dropbox.com/s/skke87vbmjjm09v/K9.par)

[Jackson]

Thanks, how do i save the K9.par as a .par file? Instead i just get k9.par as a .txt file

Edit: Nvm i just changed an existed par file to 'K9' and copied and pasted the info into that one.

[MfA_]

Ok I added the .par file to op..

[Jackson]

Ok I added the .par file to first mesage..

Thanks

[Jackson]

Mine just said 'Lithuanian' And plotted me in the middle of Poland, a little west from David's location. I'll try again just incase.

[MfA_]

Close enough

Reference populations: 1
http://abload.de/img/desktop_2013_09_15_141gkco.png

Reference populations: 10
http://abload.de/img/desktop_2013_09_15_13dushv.png


Based on the data that you have uploaded we have found that your ancestors came from Lezgins origin.


42.91% Southwest_Asian
0.99% Native_American
1.24% Northeast_Asian
36.86% Mediterranean
17.68% North_European
0.03% Southeast_Asian
0.28% Oceanian
0.00% South_African
0.00% Sub-Saharan_African

[Graham]

Mfas looks decent.. Jackson it's strange your results. I'll try it soon. To see if yours is off or not.

[Jackson]

These are the numerical results:

9.59% Southwest_Asian
1.11% Native_American
0.18% Northeast_Asian
18.54% Mediterranean
67.35% North_European
1.09% Southeast_Asian
0.86% Oceanian
0.05% South_African
1.23% Sub-Saharan_African

[Jackson]

Yeah i just get Lithuanian every time, really weird.

Edit: I'll try another family member.

Edit2: If i use 1 reference population it puts me in Lithuania, 10 puts me in Poland, and the ones inbetween in different places.

[Graham]

Aye, doesn't work for us..


Based on the data that you have uploaded we have found that your ancestors came from Lithuanians origin.

7.30% Southwest_Asian
1.33% Native_American
0.65% Northeast_Asian
14.62% Mediterranean
72.80% North_European
1.23% Southeast_Asian
1.39% Oceanian
0.67% South_African
0.01% Sub-Saharan_African

[Jackson]

Aye, doesn't work for us..

Well i guess we can compare numerical results in the meantime, but defeats the purpose. Ah well... It seemed to work for David though, maybe there is something wrong in the files we have or something?

[Graham]

100 on Med doesn't go any further than Eastern Greece, & 100 goes to Lithuania. It wont work for any Western European.

[Atlantic Islander]

I actually tried it as well, but it didn't work for me either.

K9 score
----------------------------
FINAL ADMIXTURE PROPORTIONS:
----------------------------
5.82% Southwest_Asian
0.82% Native_American
1.09% Northeast_Asian
29.84% Mediterranean
58.52% North_European
0.76% Southeast_Asian
0.63% Oceanian
0.69% South_African
1.84% Sub-Saharan_African

[MfA_]

100 on Med doesn't go any further than Eastern Greece, & 100 goes to Lithuania. It wont work for any Western European.

Lumping all Euro components into North Euro is wrong from the start if one intends to create a ''GPS'' tool.. I don't think there is much David can improve any further..

[Jackson]

I guess you can't get it that exact in northern or western Europe when there are at least 4 major components in the northern European 'block' i suppose.

[Graham]

You need an Atlantic Cluster, that pulls to the Iberian nations. Lithuania would pull anyone northeastwards from the Iberian-Atlantic cluster, there-after. The Med would pull East.

[SilverKnight]

Mozabite - North Africa (Algeria)


http://s13.postimg.org/v9n7xulqf/Untitled.png (http://postimage.org/)




based on these

9.10% Southwest_Asian
5.89% Native_American
1.11% Northeast_Asian
19.57% Mediterranean
17.56 West_ European
8.53% North_European
37.87% Sub-Saharan_African
0.02% Southeast_Asian
0.91% Oceanian
0.84% South_Africa

[Atlantic Islander]

based on these

9.10% Southwest_Asian
5.89% Native_American
1.11% Northeast_Asian
19.57% Mediterranean
17.56 West_ European
8.53% North_European
37.87% Sub-Saharan_African
0.02% Southeast_Asian
0.91% Oceanian
0.84% South_Africa

Why does yours have West_European? Is there another file?

[SilverKnight]

Why does yours have West_European? Is there another file?

Yes there is

I used k12 dodecad, it wouldn't make sense without it.

[Atlantic Islander]

Yes there is

I used k12 dodecad, it wouldn't make sense without it.

This one? (http://dodecad.blogspot.com/2011/04/k12-admixture-results-for-selected.html)

[Mark]

----------------------------
FINAL ADMIXTURE PROPORTIONS:
----------------------------
41.84% Mediterranean
39.73% North_European
15.78% Southwest_Asian
1.25% Sub-Saharan_African
0.81% Northeast_Asian
0.44% Southeast_Asian
0.15% South_African
0.01% Oceanian
0.00% Native_American
--------------------------------------------------------------------

It's obviously wrong - I tried it 6 times and I either get Germany or France.

http://i.imgur.com/5tgepHf.png

[Tropico]

I ended up in Hungary. ok...

[7eleven]

This test is crap, I ended up in Poland.

[SilverKnight]

This one? (http://dodecad.blogspot.com/2011/04/k12-admixture-results-for-selected.html)

Yeah here's mine in an illustration.

I'm DOD 240,

http://s19.postimage.org/h21r2pnk3/ADMIXTURE_DOD237_244_10.png

So far mines correlates with the 23andme global similarity map as well, maybe I did it right this time:icon_ask: it has to be done properly

[Atlantic Islander]

Yeah here's mine in an illustration.

I'm DOD 240,

http://s19.postimage.org/h21r2pnk3/ADMIXTURE_DOD237_244_10.png

So far mines correlates with the 23andme global similarity map as well, maybe I did it right this time:icon_ask: it has to be done properly

That's not what you posted though.

[SilverKnight]

That's not what you posted though.

I guess not, different one then. But still I would have gotten similar results anyways, around the area.


How about yourself?

[Atlantic Islander]

How about yourself?

I've already posted. (http://www.theapricity.com/forum/showthread.php?93834-Geographic-Population-Structure-%28GPS%29-prediction&p=1908831&viewfull=1#post1908831)

[SilverKnight]

I've already posted. (http://www.theapricity.com/forum/showthread.php?93834-Geographic-Population-Structure-%28GPS%29-prediction&p=1908831&viewfull=1#post1908831)

well.. based on yours I would say right in in between central and N. eastern France

[Mark]

well.. based on yours I would say right in in between central and N. eastern France
It places me in France and/ or Germany. It's flawed, at least for me.

[Atlantic Islander]

It places me in France and/ or Germany. It's flawed, at least for me.

Yeah, it doesn't seem to work properly.

[SilverKnight]

It places me in France and/ or Germany. It's flawed, at least for me.


Yeah, it doesn't seem to work properly.

Then why did it work for me? :confused: I'm more heterogeneous.

[Mark]

Then why did it work for me? :confused: I'm more heterogeneous.
Probably is just calibrated better toward your admixture/component ratios. But I am also heterogeneous.

[Atlantic Islander]

Then why did it work for me? :confused: I'm more heterogeneous.

How does it turn out when you actually use the K9 file?

[SilverKnight]

Probably is just calibrated better toward your admixture/component ratios. But I am also heterogeneous.

Probably, this is as accurate as I have gotten it.


How does it turn out when you actually use the K9 file?

I haven't used that one, I'm still confused on how to use it:confused:

[Mark]

I haven't used that one, I'm still confused on how to use it:confused:
Let me try and understand this... So you have only used your K11 results from GEDmatch and have NOT used the new K9 file with DIYdodecad via 'R'?

[SilverKnight]

Let me try and understand this... So you have only used your K11 results form GEDmatch and have NOT used DIYdodecad via 'R'?


Yes only k12 and k11, not the one you last mentioned, how do I use it? I tough it will work with k11 as it seems to match similarly with 23andme results

[Mark]

Yes only k12 and k11, not the one you last mentioned, how do I use it? I tough it will work with k11 as it seems to match similarly with 23andme results
There should probably be a sticky on how to do all this for those unfamiliar with Dodecad and R.
This isn't something I could detail out for you now.

[SSlava]

7.70% Southwest_Asian
0.90% Native_American
2.12% Northeast_Asian
12.39% Mediterranean
73.07% North_European
1.92% Southeast_Asian
0.92% Oceanian
0.91% South_African
0.06% Sub-Saharan_African

http://s019.radikal.ru/i602/1309/63/4b889be9f8d4t.jpg (http://radikal.ru/fp/951cc76c57ef442f923718141823272b)

[SSlava]

Reference populations to use 10
http://s019.radikal.ru/i620/1309/88/18791def4abdt.jpg (http://radikal.ru/fp/32d6a80a8cf14511b1bfcdcc4becfefd)

[Annihilus]

1

http://i631.photobucket.com/albums/uu38/ann1h1lus/1.jpg

10

http://i631.photobucket.com/albums/uu38/ann1h1lus/10-1.jpg

[Argang]

With one reference population, I'm placed in Lithuania.
Baltic Sea with three populations
Central Poland with ten.

Not too accurate, but I'm going to try this with actual Geno 2.0 results once my order is sorted out.

[Gaston]

I'm placed in West Germany. Not bad compared to others.

[Kazimiera]

1 Population
http://i1162.photobucket.com/albums/q526/katzentatzen79/Eurogenes/K9b1popGPS_zpsbdd7be40.png


According to this, I come from an open field at the edge of a forest near Strzałkowo in Poland. :D

10 Populations
http://i1162.photobucket.com/albums/q526/katzentatzen79/Eurogenes/K9b10popGPS_zps9dfa1eab.png

[MarkyMark]

Close enough

Reference populations: 1
http://abload.de/img/desktop_2013_09_15_141gkco.png

Reference populations: 10
http://abload.de/img/desktop_2013_09_15_13dushv.png




42.91% Southwest_Asian
0.99% Native_American
1.24% Northeast_Asian
36.86% Mediterranean
17.68% North_European
0.03% Southeast_Asian
0.28% Oceanian
0.00% South_African
0.00% Sub-Saharan_African

Diyabakir? Are you partially Syriac?

[Jackson]

1 Population
http://i1162.photobucket.com/albums/q526/katzentatzen79/Eurogenes/K9b1popGPS_zpsbdd7be40.png


According to this, I come from an open field at the edge of a forest near Strzałkowo in Poland. :D

10 Populations
http://i1162.photobucket.com/albums/q526/katzentatzen79/Eurogenes/K9b10popGPS_zps9dfa1eab.png

Looks like most people north of central Europe are Polish or Lithuanian now lol, i think this may be accurate for some people, but not really for northern Europeans, unless they happen to be Polish or Lithuanian.

[Atlantic Islander]

Looks like most people north of central Europe are Polish or Lithuanian now lol, i think this may be accurate for some people, but not really for northern Europeans, unless they happen to be Polish or Lithuanian.

Not accurate for Southern Europeans either.

[Patches]

10 Populations:
http://i1154.photobucket.com/albums/p535/PureEvil91/GPS10_zps21fa9950.png (http://s1154.photobucket.com/user/PureEvil91/media/GPS10_zps21fa9950.png.html)

1 Population:
http://i1154.photobucket.com/albums/p535/PureEvil91/GPS1_zps0121b5e3.png (http://s1154.photobucket.com/user/PureEvil91/media/GPS1_zps0121b5e3.png.html)

[Gauthier]

I always felt a special connection with my Lithuanian brothas. :laugh:


http://img801.imageshack.us/img801/180/x244.jpg (http://img801.imageshack.us/i/x244.jpg/)

[Vogon Poet]

Southwest_Asian 13.56%
Native_American
Northeast_Asian 1.04%
Mediterranean 33.16%
North_European 51.55%
Southeast_Asian -
Oceanian 0.48%
South_African 0.21%
Sub-Saharan_African -

Results in this:

http://i1060.photobucket.com/albums/t444/Dukh92/AutsmGeneGPS_zpse880442b.jpg

[Gauthier]

It seems like those who score higher North European than Mediterranean get automatically pulled towards Lithuania, Poland and similar places. And the people who have an almost balanced score of North and Med get taken to central Europe.

[lI]

Hahah I landed 13miles away from the actual place I grew up at:
http://imageshack.us/a/img31/8324/79je.jpg



http://imageshack.us/a/img856/8662/0hpy.jpg


It's just 3miles away from the village where my Aukstaitian grandmother was born and just 2miles away from the village where her mother, my great-grandmother, was born.


And who cares if it doesn't work for anybody else, I like this tool :naughty2:

[Argang]

I ran this test again with different numbers of reference populations, and while mostly I'm placed in Lithuania or Poland, 8 and 9 put me in Egypt, close to the Red Sea coast. This in spite of the test still saying I'm of "Lithuanian origin". Anyone else getting North African results with 8/9 populations?

[Vogon Poet]

It seems like those who score higher North European than Mediterranean get automatically pulled towards Lithuania, Poland and similar places. And the people who have an almost balanced score of North and Med get taken to central Europe.

There needs to be more diverse samples. North European is defined as Lithuanian, which means that everyone with high North European scores will drift toward Lithuania. This results in Brits, who are Northern Europeans, end up in Poland or Lithuania. The test is basically only reliable for Lithuanians and neighbour regions. Ideally there would have been sub-divisions like Iberian, Scandinavian, British, Balkan, East Slavic and so on.

[Argang]

There needs to be more diverse samples. North European is defined as Lithuanian, which means that everyone with high North European scores will drift toward Lithuania. This results in Brits, who are Northern Europeans, end up in Poland or Lithuania. The test is basically only reliable for Lithuanians and neighbour regions. Ideally there would have been sub-divisions like Iberian, Scandinavian, British, Balkan, East Slavic and so on.

You're quite right. Inserting values of Genographic Project's Russian Tatar and British reference populations yielded Lithuanian/Polish results for both. They're easily distinguishable (Tatars have 16% NE Asian, British have 0%), and the most obvious thing they have in common is North Euro as the largest component (40% for Tatar, 50% for British). Thus one shouldn't expect it to work with Geno 2.0 values.

The algorithms behind it are rather quirky. 20% Southwest Asian, 44% North European and 36% Mediterranean gets labelled Lithuanian but change those values to 20%,43%,37% and you become an Orcadian. If instead of 20% SW Asian you have 10% Northeast Asian and 10% SW Asian, the point where you become Orcadian instead of Lithuanian is 42% North Euro and 38% Med.

[SSlava]

Here are the results of man with Tartar origin seems to be (in my matches available):
Southwest_Asian 14.58%
Native_American 2.60%
Northeast_Asian 11.20%
Mediterranean 3.68%
North_European 65.94%
Southeast_Asian -
Oceanian 0.40%
South_African -
Sub-Saharan_African 1.56%

[SSlava]

But the result of the Kazakh my matches:
Southwest_Asian 18.71%
Native_American 4.27%
Northeast_Asian 47.32%
Mediterranean 2.81%
North_European 19.16%
Southeast_Asian 5.16%
Oceanian 0.91%
South_African -
Sub-Saharan_African 1.67%

[Sicilianu101]

Someone probably asked this already but how do you do this with the original files step by step? Also when do u get the GPS coordinates? Thanks!