Re: [tied] Re: Migrations and diversity in Europe (Roxana)

[Roxana] The conclusions seem to be interesting but I don't understand how
PCA works here. The PCA technique decomposes a multicomponent data field in
spatial patterns (Empirical Orthogonal Functions/EOFs) and
associated principal components (PCs) -- series that modulates the EOFs. The
PCs should be *series* of something (e.g. frequency occurrences of genetic
features), not *maps*! Are the PC series genetic fingerprints of
historically distinct populations?
------
[Davius] Yes, indeed the Principal Components (PC) seems interpretable as
fingerprints of at least three people migration in Europe: Neolithical,
Uralic and IE.

[Roxana] How many genes and spatial points were used in this PCA? (We should
know these numbers in order to assess the statistical significance of the
results.)
---------
[Davius] The study was carried out comparing the frecuency of 39 genes (in
the '70) and in the '90 was improved using 95 genes. Thus the covariance
matrix is 95x95.
The percentage of variance explained considereing PCI are::
-PC1 --------------28%
-PC2 --------------22%
-PC3 --------------24%
- PC4--------------11%

Thus the cumulative percentage of genetic diversity in Europe of Neolithic,
Uralic, IE and Greek expansions is 85% of the total variance!!

Davius Sanctex