[gmx-users] Setup box dimensions

[Nuno R. L. Ferreira]

Hi all.

One doubt that puzzles me a bit, is setting up the box dimensions.
Read  in one of J. Kerrigan's tutorials, that ideally we should set -d at no
less than 0.5 nm to an upper limit of 0.8 nm.
But water only tends to have uncorrelated motions after about 0.9 nm, at
room temperature. So, it seems that the minimum distance should be 0.9, in
such a way that at the edge of the box, water should "see" disordered water.
If the protein is globular, this roughly means that between two periodic
images (-d = 0.9) of the system, the protein (ab) is 1.8 nm apart.

                    ---------------------------
                   |                    |                    |
                   |        ab        |        ab         |
                   |                    |                    |
                    ---------------------------
                              ^       ^
                               |        |
                               <0.9>       ^
                                                |
                               <    1.8    >


If rlist = 1.0 , rvdw = 1.4 , rcoulomb = 1.0 (PME), does this mean that the
short-range non-bonded interactions (LJ & coulombolic) of atom_b sees 0.1nm
of rlist of atom_a?  Does it influence the stability of the protein?
McCammon reported in 2000 (JPC B,104,3668), that box size strongly
influences the stability of a peptide.

In simple words, what is a safe value for the box size, to prevent these
kind of problems? And how to play with -d (editconf),rlist,rcoulomb and
rvdw?
If I increase any of these 3 last params, do I have also to increase the box
size?

Regards,
NUno

P.S. I've been doing some MD runs with -d = 1.4, and the treatment of
short/long range forces with values described earlier (last e-mail in the
list).
        My aim is to see the conformational stability of an alpha-helix, so
I'm interested to know if I can decrease my box size in order to save
computational time, without interfering with the protein stability.

######################################
Nuno Ricardo Santos Loureiro da Silva Ferreira
Departamento de Química
Faculdade de Ciências e Tecnologia
Universidade de Coimbra
3004-535 Coimbra - Portugal
Fax: +351 239 827703  - www.biolchem.qui.uc.pt
#######################################