CHARMM Development Project Forums User Discussion & Questions General Chemistry Discussions comparing groups of molecules

I have a bunch of different configurations of a single molecule under some sort of enviromental conditions. They are not of course the same configuration because of fluctations.

I have another bunch of different configurations of the same molecule under different or the same conditions (this is unknown). How would I tell if they are under the same conditions or not?

This seems like a very general statistical question that I am sure some one has dealt with before in some form. Does any one know? I need some good search words to google and get started...

Distributions of torsions and properties such as energy should be similar if either the conditions are the same, or the differences in the conditions don't have much impact.

What I was thinking was more of looking at the individual types of configurations by some sort of clustering algorithm. So for instance, if 50% of the configurations belong to cluster A and 50% of the configurations belong to cluster B for both of the "different" conditions then the conditions are not different.

There is torsion-aware clustering code in CORREL; the clusters will, of course, depend on the underlying distributions. For smaller molecules, comparing the actual distributions may be better.

Re-evaluating the energy of both sets in a consistent way should also be performed; significant differences in the distributions would mean they are not equivalent.