[adegenet-forum] DAPC & Ploidylevel

Stefanie Hilpert hilpert at ipk-gatersleben.de
Wed Oct 29 11:24:26 CET 2014


I've already asked the question a week ago, but I'll just try again, so here we go:

Dear everybody,

I am currently using the adegenet package to perform a structure analysis of my microsatellite dataset and compare it to the results of an analysis using STRUCTURE software. The organism I am working on is an apomictic plant and I am aware that STRUCTURE is probably not adequate because it assumes HWE and asexuality violates HWE. Nevertheless we use STRUCTURE analysis for apomicts, because in most of the cases the assigned number of groups correlate to biological traits.  Knowing that there is a bias using STRUCTURE we decided to perform a DAPC additionally.
But now I ran into another problem using adegenet. I am working with a mixed ploidy system with ploidies ranging from 4 to 11. To implement the data into the adgenet package we coded all individuals as 11x because otherwise it was not possible to load the data. Now I am wondering how big is the bias if the calculation assumes that for example a tetraploid is now a hendecaploid and if I could still trust the results. I am asking because the results of the DAPC are completely different to the ones of STRUCTURE which puzzles me a bit because I somehow at least expected correlations (the number of optimal k is the same, but the assigned individuals to the clusters differ completely).

I would appreciate some help

Stefanie Hilpert



------------------------------------------------------------------------------
Stefanie Hilpert
-PhD Candidate-
Dept. of Cytogenetics and Genome Analysis
Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)
Corrensstraße 3, D-06466 Gatersleben Germany
+49 (0)39482 5673

IPK Graduate School
International Max-Planck Research School


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20141029/59d694c1/attachment.html>


More information about the adegenet-forum mailing list