[adegenet-forum] double-zero problem in genetic data

[Xavier Turon]

Hello,

I'm new in this forum and I have just done my first DAPC analyses. I 
have a question, and sorry if it is naive or has already been discussed 
at previous times.

Coming from the field of ecology, I am reluctant to analyse matrices 
with many zeroes with PCA. This is typical of matrices of species 
abundance, where absence of a species in two localities (double-zeros) 
does not mean at all that both localities are related. Many double 
absences can result in spurious high covariances.

I wonder if the same cannot be said of the individuals * alleles 
matrices, which by their very nature have plenty of zeros (i.e., the 
absence of a given allele in two individuals can be interpreted as 
indicating similarity between them?)

Thanks for your inputs on this matter,



Xavier
---

Xavier Turon
Dept. of Marine Ecology
Center for Advanced Studies of Blanes (CEAB, CSIC)
Accés a la Cala S. Francesc 14
17300 Blanes (Girona)
Spain

E-mail: xturon at ceab.csic.es
Phone: +34 972 336101
Fax: +34 972 337806
http://www.ceab.csic.es/
http://www.ub.edu/beb/english/invertebrates/turon/turon.htm