[gmx-users] Calculating bond energy

[Mark Abraham]

Yulia Einav wrote:
> Thank you for your reply, Mark.
> My protein has four disulfide bonds. I did in silico mutation of one of them and, then, MD simulations. Now I want to know whether breaking one bond influence the others. I can measure the distance between the centers of mass of two sulfur atoms, but is it sufficient to supply the information about bond strength of each disulfide? I guess I mean the bond dissociation energy (not the average one, but rather for each bond individually) of the remaining S-S bonds in the protein.

In principle, I guess you could measure such a bond dissociation free 
energy with free energy perturbation (see 
http://en.wikipedia.org/wiki/Free_energy_perturbation in the first 
instance) but offhand I'd guess that a state with two bound sulfur atoms 
is going to be too far removed from two -SH groups that are free to 
dissociate for FEP methods to work in finite time.

It's definitely true that any S-S atom distance doesn't measure anything 
relevant. Even if you haven't constrained your bond lengths, there's no 
reason to suppose that the contribution to the total potential energy of 
the "bond stretching" term has any strong relationship with bond strain 
or such.

This is a good lesson in why one needs to design a computational study 
with a specific goal in mind, and then to choose tools that can do the 
job - e.g. measure that observable.

Mark