Friday, February 9, 2024

When the tools and settings matters. Part 1 and Part 2

 When the tools and settings matters. Part 1

Here I present two models with exactly same source and target populations.
The first one is unscaled, and the second one is scaled.
As You can see the most significant change affect the EHG level which is 2.5 times higher in the unscaled.
You may wonder which one is the most correct. Well, the unscaled coordinates are the one that are produced by the smartpca in their unaltered form.
While the scaled coordinates are deliberately altered by an algorithm. I don't know exactly how this algorithm works.
But the main idea of scaling is too artificiality increase the distance for distant and drifted populations. For instance, in the East Asian populations but also the local drifted pops like CHG and Turkey_N. Those artificial changes alter the result in different ways. As You can see for Armenians it is creating a masking effect that hides the real EHG.
It must be noted the qpadm numbers are close to the unscaled numbers. But qpadm also has its own tips and secrets.
In any case questions like how much EHG, Levant or CHG do have an X population do not have a strict answer. They are rather ranges than just one correct answer.

When the tools and settings matters. Part 2

Here I add a modification to the models present in the first part. I am adding samples from South Caucasian farmers. Similar populations existed all over historic Armenia and north Mesopotamia. Here we use a generic term for them as Central farmers.
The addition of this new population changes once more the results in both unscaled (the first) and the scaled (the second). The EHG increases once more (average 5.5) but the difference between unscaled and scaled is now very low. You can also notice that the fits are improved, and the errors are reduced. Which means that those farmers are a better source for Armenians (and not only Armenians) than those four distal pops.
The reason of this is that modern Armenians and not only Armenians are not a real mixture of four West Asian distal populations.
Central farmers =//= Turkey_N + Kotias + Ganj Dareh + Natufian
Central farmers are only approximately similar to the addition to those four populations. But the real history of West Asia, which is not fully uncovered, most probably will show that the deep origins of Central farmers is different. That is something of the future but at this stage we must know that Central farmers are more basal than the four distal pops combined. The Turkey_N and the Kotias-CHG have a WHG related alleles which acts as a mask for the EHG. Thus, the addition of a basal population like Central farmers lacking this WHG causes an increase of EHG. While the difference between the scaled and the unscaled decrease because the model does not rely anymore on distal four populations whose coordinates are artificially changed.

No comments:

Post a Comment