My Global25 results, Northern Greek.

[dosas]

Here are my recent Global25 results:


[1] "distance%=2.6831"

TP_scaled

Armenia_EBA,39.8
Barcin_N,39
Yamnaya_Samara,18.8
Jarawa,1
Yoruba,1
WHG,0.4

Also G25 euro (I am the black dot):

<a href="http://s264.photobucket.com/user/theopard/media/G25_Europe_scaled_zps5wewaijc.png.html" target="_blank"><img src="http://i264.photobucket.com/albums/ii181/theopard/G25_Europe_scaled_zps5wewaijc.png" border="0" alt=" photo G25_Europe_scaled_zps5wewaijc.png"/></a>

My original Myheritage/GEDMATCH thread (https://www.theapricity.com/forum/showthread.php?262382-Can-you-help-me-understand-my-GEDmatch-results-Northern-Greek)

[Kaspias]

Where are you from? Thrace or Thessaloniki?

[dosas]

Greek Thracian (paternal side) from Ivaylovgrad. You commented on my original thread I think.

[Kaspias]

Greek Thracian (paternal side) from Ivaylovgrad. You commented on my original thread I think.

Ya i remembered. I just got my global25 coordinates, we can compare our results after i run them.

Did you try modern spreadsheet?

[dosas]

To be fair, mate, the dude sent me the stuff in an email. I haven't done anything on my own because I am kinda of a newb, so I am open to suggestions.

[Lucas]

Here are my recent Global25 results:


[1] "distance%=2.6831"

TP_scaled

Armenia_EBA,39.8
Barcin_N,39
Yamnaya_Samara,18.8
Jarawa,1
Yoruba,1
WHG,0.4

Also G25 euro (I am the black dot):

<a href="http://s264.photobucket.com/user/theopard/media/G25_Europe_scaled_zps5wewaijc.png.html" target="_blank"><img src="http://i264.photobucket.com/albums/ii181/theopard/G25_Europe_scaled_zps5wewaijc.png" border="0" alt=" photo G25_Europe_scaled_zps5wewaijc.png"/></a>

My original Myheritage/GEDMATCH thread (https://www.theapricity.com/forum/showthread.php?262382-Can-you-help-me-understand-my-GEDmatch-results-Northern-Greek)

You are very south on PCA for northern Greek. You are really 100% Thracian? You got PCA directly from Davidski?

[Sikeliot]

You are very south on PCA for northern Greek. You are really 100% Thracian? You got PCA directly from Davidski?

He has Pontic ancestry as well. Either way he is somewhere between Cretans, south Italians ,and mainland Greeks and doesn't really plot exactly with any group.

[FilhoV]

Nice results

[Vasconcelos]

Uncommon and cool, I'm also a bit of an outlier myself. When you run your data through Cluster Analysis, who do you cluster with and which are the shortest distance samples (which might not be the ones who look closer on the PCA)?

[dosas]

Uncommon and cool, I'm also a bit of an outlier myself. When you run your data through Cluster Analysis, who do you cluster with and which are the shortest distance samples (which might not be the ones who look closer on the PCA)?

Thank you. I don't know how to check that.

[Papastratosels26]

Great results man. :[emoji106]

Στάλθηκε από το G3311 μου χρησιμοποιώντας Tapatalk

[Vasconcelos]

Thank you. I don't know how to check that.

Download PAST 3, open your Global 25 file, select all samples and click on multivariate > clustering > classical.
This will give you the dendrogram (https://en.wikipedia.org/wiki/Dendrogram) to see who's in your cluster (for example I get a couple of guys form Leon y Castilla and one from Castilla La Mancha on my branch)


You can also go: multivariate > similarity and distance indices.
This will give you a matrix of distances each samples has to the other. It will be pretty massive if you select all samples, so you can just delete all those that are not relevant, since the result will not change (unlike in a PCA plot). I advise copying the results to an Excel sheet and calculating the minimum value of your row, then finding out which sample it corresponds to. As expected those of your cluster will generally be amongst the closest to you.

[Levant15]

Download PAST 3, open your Global 25 file, select all samples and click on multivariate > clustering > classical.
This will give you the dendrogram (https://en.wikipedia.org/wiki/Dendrogram) to see who's in your cluster (for example I get a couple of guys form Leon y Castilla and one from Castilla La Mancha on my branch)


You can also go: multivariate > similarity and distance indices.
This will give you a matrix of distances each samples has to the other. It will be pretty massive if you select all samples, so you can just delete all those that are not relevant, since the result will not change (unlike in a PCA plot). I advise copying the results to an Excel sheet and calculating the minimum value of your row, then finding out which sample it corresponds to. As expected those of your cluster will generally be amongst the closest to you.

Just out of curiosity, which index should be used in Task3? Euclidean? Or maybe Correlation?

Sent from my FRD-L04 using Tapatalk

[dosas]

I can't seem to open my .dat file with PAST 3, it gives me a blank window.

Nvm, done it, now it says bitmap size too big error.

[Vasconcelos]

Just out of curiosity, which index should be used in Task3? Euclidean? Or maybe Correlation?

Sent from my FRD-L04 using Tapatalk

I can't give you a proper justified answer unfortunetly, but I generally leave it at Euclidean (which is the linear distance between each sample)

When I set to Correlation my two most similar samples remain the same (HG01500 and HG01501), while the third changes from a Manchego to a Catalan. Most of the following are also from Spain, although a few form Portugal also pop up. I get a few Basques before some Portuguese samples (indeed my correlation average with Pais Vasco is higher than the Portuguese..oh well).
This isn't surprising, the trend also happens with Euclidean distances, I just have wierd results.








I can't seem to open my .dat file with PAST 3, it gives me a blank window.

Nvm, done it, now it says bitmap size too big error.

Delete a few samples, like Austrians, Portuguese, Germans and so on. They don't really matter for you when checking clusters, correlations or distances.

[FilhoV]

Download PAST 3, open your Global 25 file, select all samples and click on multivariate > clustering > classical.
This will give you the dendrogram (https://en.wikipedia.org/wiki/Dendrogram) to see who's in your cluster (for example I get a couple of guys form Leon y Castilla and one from Castilla La Mancha on my branch)
Hey Vasconcelos if I send you my coordinates would you mind running it through that sample?

You can also go: multivariate > similarity and distance indices.
This will give you a matrix of distances each samples has to the other. It will be pretty massive if you select all samples, so you can just delete all those that are not relevant, since the result will not change (unlike in a PCA plot). I advise copying the results to an Excel sheet and calculating the minimum value of your row, then finding out which sample it corresponds to. As expected those of your cluster will generally be amongst the closest to you.

If I give you my coordinates could you run it through that spreadsheet?

[Vasconcelos]

If I give you my coordinates could you run it through that spreadsheet?

Yeah, I guess so. It's really easy to do these things with PAST 3 anyway, it's just a bit boring to do the excel thing because there are so many samples

[Kaspias]

I ran nMonte with your populations, seems like it didn't work for me.(See:distance) Expected, we have different ancestry just same location.

[1] "distance%=8.098 / distance=0.08098"

Kaspias_scaled

Barcin_N:I0707 41.2
Yamnaya_Samara:I0370 40.4
Armenia_EBA:I1633 10.7
Jarawa:JW-10 7.7

[J. Ketch]

I ran nMonte with your populations, seems like it didn't work for me.(See:distance) Expected, we have different ancestry just same location.

[1] "distance%=8.098 / distance=0.08098"

Kaspias_scaled

Barcin_N:I0707 41.2
Yamnaya_Samara:I0370 40.4
Armenia_EBA:I1633 10.7
Jarawa:JW-10 7.7
Those are individuals. You want to use the population averages sheet.

[Aren]

Post your coordinates here and I'll do some runs for you.

[dosas]

The Eastern Rumelia Greek Thracians, afaik, were a very segregated group in relation to Bulgarians and Turks.

[Kaspias]

Those are individuals. You want to use the population averages sheet.

I don't think there will be so much difference. Thank you!

[dosas]

Download PAST 3, open your Global 25 file, select all samples and click on multivariate > clustering > classical.
This will give you the dendrogram (https://en.wikipedia.org/wiki/Dendrogram) to see who's in your cluster (for example I get a couple of guys form Leon y Castilla and one from Castilla La Mancha on my branch)


You can also go: multivariate > similarity and distance indices.
This will give you a matrix of distances each samples has to the other. It will be pretty massive if you select all samples, so you can just delete all those that are not relevant, since the result will not change (unlike in a PCA plot). I advise copying the results to an Excel sheet and calculating the minimum value of your row, then finding out which sample it corresponds to. As expected those of your cluster will generally be amongst the closest to you.

I've managed to do it and my cluster seems similar to the graph, various Greek, Greek Cretan and South Italy.

[Kaspias]

The Eastern Rumelia Greek Thracians, afaik, were a very segregated group in relation to Bulgarians and Turks.

Yes, on the other hand Turks and Bulgarians usually mixed and lived together.

[FilhoV]

J.Ketch was nice enough to use the entire spreadsheet and this is what I got

[1] "1. CLOSEST SINGLE ITEM DISTANCE%"
Spanish_Andalucia Spanish_Extremadura
2.331287 2.571400
Portuguese Spanish_Murcia
2.708195 2.775317
Spanish_Castilla_Y_Leon Spanish_Canarias
3.001261 3.152721
Spanish_Castilla_La_Mancha Spanish_Baleares
3.156557 3.213796

[1] "distance%=0.959"

FilhoV_scaled

Ukraine_N_o,18.8
Wales_CA_EBA,14.4
Spanish_Andalucia,10.2
Wales_N,10.2
Petrovka_MLBA,10
Vucedol,7.2
Spanish_Canarias,7
Maros,6.2
Tisza_LN,4.4
CHG,3.6
Comb_Ceramic,2.8
Iberomaurusian,2
Yoruba,1.6
Mozabite,0.6
Algerian,0.4
Biaka,0.2
Romania_HG,0.2
Saharawi,0.2