[gmx-users] analysis of different groups with g_energy

[Mark Abraham]

Peyman Yamin wrote:
> Thanks! 
> 
> I had somewhere in my mind a memory of seeing an interactive g_energy some 
> time ago, but it could be another version of it or that place in my mind 
> would need a defragmentation!  comments helped indeed anyway.
> 
> If group X is one molecule, you would know from it's energy, i.e. the sum of 
> interactions its atoms have with themself and with the surrounding solvent, a 
> plenty of useful things. It could, if analysed with an eye on physicochemical 
> nature of interactions and the resulting conformers, could give neat insights 
> in some cases. e.g. how would one find some attractors about a certain 
> arbitrarily defined size-related property, in terms of stability? I thought 
> if molecule X is tending to have some conformer with the least energy in 
> medium A, it would also be tending to have an alike conformer in some other 
> medium of the same nature to the first one.

In general, you can't just ignore entropic considerations here - that's 
one reason for doing condensed-phase MD, rather than vacuum calculations 
of some sort. MM forcefields are typically parameterized to reproduce 
certain physical or quantum chemical computational observables. Unless 
you can show a direct relationship between some of these and your 
property of interest, then you're on very shaky ground using the MM 
forcefield.

> This is what I think, and, you might want to correct me of course.
> 
> But what I don't know now is if the tot-energy output from g_energy when one 
> chooses a certain molecule, is actually what I think it is, or it's 
> calculated in another way? How would I know?

"Total energy" in g_energy and the .log file is for the whole system. An 
mdrun -rerun with a topology that's had most of its contributions 
zero-ed out is one way to get your single-molecule total energy.

Mark