From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Sat Oct 31 2009 - 14:23:13 CDT

On Tue, 2009-10-27 at 18:12 +0200, politr_at_huji.ac.il wrote:
> Dear all who answered to me.

dear regina,

> I'm glad that my question raised such a lively discussion,
> and many thanks for sharing your experience with me.
> I have already performed simulation with TIP3P and now I
> want to see how sensitive are my results to the water
> model (e.g., SPC/E) and compare between these.

with the same reasoning, you could also do a comparison
with different force fields for the peptide _and_ the
water.

> I'm working with 16 residue peptide that has a hairpin
> conformation (seen experimentally). Do you think that
> it is useless to compare between the effects? Regina

well, it is a really difficult question. of course,
you can extract some effect of the strength of water
peptide interactions on structure, but since you at
the same time massively change the balance between
peptide and water, it is in my opinion difficult to
come to easily identifiable conclusions. if anything,
i would follow a perturbation theory approach, i.e. do
very small changes to the water potential and look for
the response to it. those calculations are, however,
likely to converge very slowly.

i think a more straightforward approach would be to
completely swap the force field, e.g., to gromos, amber,
opls, ... and see how those behave. at least this is
what people that i have been working with done in the
past. each of the force fields has its preferences and
its idiosyncrasies. e.g., in charmm dihedrals are generally
quite "stiff" (which helps to keep the conformational phase
space small and simulations converge faster) and in gromos
they tend to be very "sloppy", which lets it find configurations
that charmm will never get into. others are somewhere in
between, and they keep changing with revisions (esp. amber).

since you change many parameters at the same time, the
results will also be difficult to interpret in detail,
but you have consistency between water (i.e. solvation effects)
and peptide.

cheers,
    axel.

-- 
Dr. Axel Kohlmeyer  akohlmey_at_gmail.com 
Institute for Computational Molecular Science
College of Science and Technology
Temple University, Philadelphia PA, USA.