most discriminating loci?

[gordon]

hi, first off, please forgive me if this question seems quite naive,
I'm afraid to say I'm relatively new to this particular field, so the
answer could be quite obvious. I've just come across a particular
problem, and would like to try to use the structure program to try and
solve it, if it's possible.

The problem is feature selection. I've been given a set of 44
replicated samples (so 88 in total), with allelelic information given
for 30 different loci. I've been asked to determine which of the loci
are the most discriminating. In other words, my task is to extract a
hopefully small subset of the loci that explains the variation between
all the samples. Does Structure implement anything that may be useful
for this?

In my mind the solution should be similar to principal components
analysis (PCA). Unfortunately, I don't know how to get this data into
a format that I could use with PCA.