[adegenet-forum] 3D sPCA and the use of temporal data

[Jombart, Thibaut]

Dear Sibelle, 

yes, I think it makes sense, although you'll probably have a harder time interpreting and plotting the results. 

But basically, all you need to do is formulate the spatio-temporal proximities in a proximity matrix (terms >= 0, diagonal =0). Another, more complicated approach would be discretizing the temporal data and have say 'T' time steps with one matrix of proxity each. You'll then have to coordinate 'T' sPCA, which is doable using K-table approaches such as multiple co-inertia or STATIS (all in ade4), but substantially more of a pain.

Cheers

Thibaut




________________________________________
From: adegenet-forum-bounces at lists.r-forge.r-project.org [adegenet-forum-bounces at lists.r-forge.r-project.org] on behalf of Sibelle Vilaça [sibelletorres at gmail.com]
Sent: 13 February 2013 14:37
To: adegenet-forum at lists.r-forge.r-project.org
Subject: [adegenet-forum] 3D sPCA and the use of temporal data

Hi all,
I'm working with a domestic species and I have been trying to integrate spatio-temporal data with haplotype frequencies. I've been working with some ancient DNA data, associated with spatial location. I have a small fragment of the mitochondrial DNA for several hundreds of individuals, each associated with a geographical coordinate (x,y) and to a 14C dating. I can see that there is a strong geographical correlation (geographically close samples show related DNA haplotypes in similar frequencies), but I can also see that this distribution is also strongly correlated with sampling time. Because I'm working with a domestic species, this strong influence of time has to do with Neolithic migrations, the change of haplotypes frequencies is directly correlated with human migrations and their time of arrival in different European locations.

Following the list, a saw it was possible to perform a 3D sPCA using depth data. So, I was wondering if instead of depth data as a Z coordinate, it would be possible to integrate temporal data on a sPCA, using something like (x, y, time)? I have a radiocarbon dating for each sample, but I was thinking of using time frames (or categories) related to the different human cultures (Neolithic, Mesolithic, etc) since the change in haplotype frequencies is directly related to the changes in human culture....

Is it possible?

Sibelle