Sicily and Myceneans on a PCA

Francesco

Regular Member
Messages
303
Reaction score
267
Points
63
Ethnic group
Italian (tuscan)
I'm pretty new to population genetic and maybe my question could sound a bit naive, so I apologize in advance if I'm asking some nonsense.
Anyway, I've noticed how some PCAs, expecially the ones coming from academic sources, tend to represent Sicily to the west of Myceneans, while other PCAs have sicilians a little bit more shifted towards Anatolia/Caucasus.
I'm wondering if there is a reason behind.
 
Are we talking about shifting Sicily to the Caucasus AHEAD OF MYCENEANS???
 
Yes. That's seems strange to me. Unfortunately, I still have too few posts to share links or image, but in some PCAs you have sicilians shifted in regards to Mycenean toward central Italy, while in other PCAs you have them more shifted towards Caucasus.
 
Since I can finally post links and images, I'll shares the following PCAs as an example. In the first two PCAs, wich are taken from academic papers, Sicilians are clearly "west" of Mycenaeans, shifted towards the western mediterranean and northern central Italy. On the third one, on the other hand, they are a bit shiftd toward Anatolia and the Caucasus. I wonder if there could be a reason behind this difference.

1) https://i.imgur.com/A0f7WF8.png

2) https://i.imgur.com/B6FzxkY.jpg

3) https://i.imgur.com/2xPE1Cb.png
 
Since I can finally post links and images, I'll shares the following PCAs as an example. In the first two PCAs, wich are taken from academic papers, Sicilians are clearly "west" of Mycenaeans, shifted towards the western mediterranean and northern central Italy. On the third one, on the other hand, they are a bit shiftd toward Anatolia and the Caucasus. I wonder if there could be a reason behind this difference.

1) https://i.imgur.com/A0f7WF8.png

2) https://i.imgur.com/B6FzxkY.jpg

3) https://i.imgur.com/2xPE1Cb.png

The third PCA is done by Anthrogenica users with "south Italian" samples on G25, so you have your answer there.
On the other hand, if you wonder about PCAs in which Myceneans are more "west", that is more downshifted on the Y axis than modern Sicilians, whereas in others they seem to overlap with Sicilians, I guess it depends on the methods used and maybe something to do with projectional bias, but in every PCA Sicilians and Myceneans are equally "MENA" shifted, that is rightshifted on the X axis. Ancient Greeks too seem to have been on the same X value as modern Sicilians.
MExlHyN.png


aneli2021-fig1c.png
 
I'm pretty new to population genetic and maybe my question could sound a bit naive, so I apologize in advance if I'm asking some nonsense.
Anyway, I've noticed how some PCAs, expecially the ones coming from academic sources, tend to represent Sicily to the west of Myceneans, while other PCAs have sicilians a little bit more shifted towards Anatolia/Caucasus.
I'm wondering if there is a reason behind.

Academic PCAs usually use few samples, because their aim is to give the general picture rather than detail on the posistion of a single population. Apart from specific PCAs. In general, in my opinion PCAs should not be interpreted as if they were geographical maps. It is true that genetics follows geography, but the orientation of PCAs can change for many reasons even if the genetic positions of the samples represented in the PCAs do not actually change. In a PCA, the position of a sample can change depending on the samples you fill in, because the samples in a PCA are like interacting with each other.


Generally speaking, Myceneans have higher EEF than modern populations, including Sicilians. On the other hand, the Mycenaeans were a Bronze Age population and the ethnogenesis of the Greeks was not quite completed. The position of a single population is truly accurate when as many possible samples are included in a PCA.

Using all academic samples with Dodecad K12b. The trends are these. Sicilians are represented by all possible samples: Western Sicily, Central Sicily, Eastern Sicily, Trapani and Siracusa. Since Dodecad K12b was created before the 'discovery' of ancestral components, the position of some ancient DNA samples may not be 100% accurate.


qSxCdAd.png


LrtrZW7.png


murxg9v.png



n7V4OTB.png


ed9uxP6.png
 
In a PCA, the position of a sample can change depending on the samples you fill in, because the samples in a PCA are like interacting with each other.

I wasn't pretty aware of this, thanks


Using all academic samples with Dodecad K12b. The trends are these. Sicilians are represented by all possible samples: Western Sicily, Central Sicily, Eastern Sicily, Trapani and Siracusa. Since Dodecad K12b was created before the 'discovery' of ancestral components, the position of some ancient DNA samples may not be 100% accurate.

So, if I've understood properly, more recent, academic PCAs should be held as more reliable as long as the position of ancient populations in relation to modern samples is concerned, right?
 
So, if I've understood properly, more recent, academic PCAs should be held as more reliable as long as the position of ancient populations in relation to modern samples is concerned, right?

The problem is not so much whether they are more or less recent (although it is true that in more recent ones there may be more samples because many samples are analysed more recently).

I meant that if you want to see the exact position of a modern population in a PCA in relation to modern and ancient populations, you have to include all the modern populations surrounding this population in a PCA.

In the case of Italy, if you want to see the exact position of a modern regional population, you have to put in the PCA at least the samples of all the other 20 Italian regional populations, and of all the populations that genetically border Italians (Spanish, Portuguese, Southern French, Swiss Italian, Swiss French, Greek, Albanian, Bulgarian, Romanian, etc.). Furthermore, all populations should have a similar number of samples (and this is not always the reality even in academic PCAs).
 
Thanks for the explanation. I'm a bit confused though: since we have the four ancestral polulations (WHG - ANF - EHG - CHG) in the same position in the PCA, roughly at the four corner of it, why do we need to add modern surrounding populations as well? shouldn't those four ancestral component be enough to have the exat position of a modern (or ancient) polulation on a PCA?
 
Thanks for the explanation. I'm a bit confused though: since we have the four ancestral polulations (WHG - ANF - EHG - CHG) in the same position in the PCA, roughly at the four corner of it, why do we need to add modern surrounding populations as well? shouldn't those four ancestral component be enough to have the exat position of a modern (or ancient) polulation on a PCA?


For reasons I have already explained to you. More populations are added more information you get. It is clear that ancestral components play a role (they are more than four, though) but the majority of PCAs do not rely directly on them. On Eupedia we use, for example, Dodecad K12b and the 12 components of K12b do not correspond exactly to the ancestral components even though K12b's components have some kinda of relation with the ancestral components.

Maybe I can't explain myself and we can't understand each other. Let me give a practical example.

Which of the PCAs do you find most informative?


1) only 1 individual per region

GyorKab.png





2) a small group of individuals per region

P44sZ5M.png



3) almost all Italian regions covered

vi7si0Q.png



4) Non-Italian samples added to the PCA

A27uK8n.png


AKtAQls.png
 
Maybe I can't explain myself and we can't understand each other.

Absolutely not, you're very clear (and patient) in your explanations. The fact is that I'm completely new to this field of knowledge and I'm not quite at ease with communicating in english either. So I probably seem more slow than I really am:embarassed:

Going back to the topic, I can understand that the last PCAs, the ones with more samples, are clearly more informative. The thing that keeps eluding me, onthe other hand, is the following.
Let's take, for example, the one Veneto sample from the first PCA (Veneto_Alp022): does its absolute position change if we put it in one of the last PCAs you posted?
You probably already answered to this interrogative, so I apologize if I still didn't get it.
 

This thread has been viewed 2967 times.

Back
Top