I've performed many umbrella sampling and meta-dynamics simulations to report the interaction energies (free energy of association, for example) of proteins. However, I am curious as to how useful the potential energy values generated by Gromacs could be? If I record the rotational motion of protein A against protein B (part of a dimer) and report how the potential energy changes as a function of rotation, how, useful is this?
The problem with total potential energy in such a case is that for a reasonably large system, it will fluctuate a lot. And even though you could in principle obtain a full thermodynamical description of your process (delta G from umbrella sampling, delta H from averaging the total energy, and delta S from simple algebra), the statistical noise will probably be an order of magnitude larger than the change itself.
You can, however, record the changes in selected pairwise interaction energies, and these may be useful to see (or show) what's going on in the simulation, provided that you have a specific question to ask. Just keep in mind that these energies are often counterbalanced by opposite changes in interactions somewhere else, so they need not determine the behavior of your system.
The problem with total potential energy in such a case is that for a reasonably large system, it will fluctuate a lot. And even though you could in principle obtain a full thermodynamical description of your process (delta G from umbrella sampling, delta H from averaging the total energy, and delta S from simple algebra), the statistical noise will probably be an order of magnitude larger than the change itself.
You can, however, record the changes in selected pairwise interaction energies, and these may be useful to see (or show) what's going on in the simulation, provided that you have a specific question to ask. Just keep in mind that these energies are often counterbalanced by opposite changes in interactions somewhere else, so they need not determine the behavior of your system.
Dear Anthony,
As Milosz indicated, using the approach you mentioned will not yield to accurate results. I suggest that you take the important cores in both proteins (exclude all parts of protiens which you expect that will not contribute to the interaction) and conduct a rotation using the semi-empirical method AM1 (uo t0 100 atoms) or even DFT (up to 60 atoms).
These calculations will give you a good picture on the interaction energy in terms of enthalpic, entropic and free energies.
Hoping this will be helpful,
Rafik
- Badges
- Science topic