[gmx-users] bond deviates....

K.A. Feenstra Feenstra at chem.vu.nl
Tue Sep 17 17:12:32 CEST 2002

MOLTENI at unisi.it wrote:
> Hi
> shortly after the beginning of a MD simulation in DMSO I get some
> warnings like "pressure scaling more than 1%" and, after that, an error
> "bond deviates more than half its own length" (after which the
> simulation stops), and also the temperature seems to explode.
> Does anyone have an idea as to what can be the reason?
> Can it be due to the fact that I'm bringing the system directly from "0
> K" (energy minimization) to 300 K? Is it (more) advisable to do short MD
> runs at, let's say, 100 K --> 200 K and finally the "long" simulation at
> 300 K?
> A couple of other questions:
> -  apart from the various changes in the topology file some of you
> already explained me about (thanks, by the way), is there anything else
> that has to be changed when doing MD in DMSO (instead of water)? For
> instance, what about the compressibility (of water) that is used by
> default? Should it be changed, and where?

IIRC compressibility of DMSO is slightly different than for water,
but since it only affects the effective scaling time constant (tau_p)
it is not very important (i.e. anything within an order of magnitude
should do). If you want to change it, it is in the .mdp file.

> - Is it really useful (and why) to do a short MD with position
> restraints on the protein, peptide.. before the "real" simulation?
> Also if I already energy minimized the system (protein + solvent)? I
> know there are no "universal recipes" on the scheme to follow (minimize
> first the protein in vacuo? minimize protein and solvent together? bring
> directly or stepwise to target temperature?...) but.... does any of you
> have one? ;-)

Yes, it is. You want to equilibrate the (newly generated) protein-solvent
interface without distorting your protein. There can be bad contacts and/
or holes between protein and solvent. If you start a free simulation with
these holes, they will 'collapse' potentially distorting the protein 
locally. Energy minimization by itself will not be able to close these

The scheme I use routinely (and only rarely lets me down) is:
1) em protein in vacuum
2) solvate
3) em solvated system
4) md w/ posres on protein (and ligands etc.)
5) 'production md'
If you want to be *extremely* careful of your starting structure (e.g.
you are optimizing a model), you could even first em with posres,
and you could do several steps of md/posres where successively more
atoms are 'set free', e.g. first only hydrogens move, then only 
sidechains, then all but Ca.

As for initial temperature, I simply generate 300K velocities at the
start of the md/posres run. The production run usually is a simple
continuation, although I also often re-generate velocities several
times (with different seeds) to do some more or less independent
runs of the same system. I have never been able to detect artefacts
from this, although it might give rise to some distortions.


 ________ ___________________________________________________________
|        | Anton Feenstra                                            |
| .      | Dept. of Pharmacochemistry - Vrije Universiteit Amsterdam |
| |----  | De Boelelaan 1083 - 1081 HV Amsterdam - The Netherlands   |
| |----  | Tel: +31 20 44 47608 - Fax: +31 20 44 47610               |
| ' __   | Feenstra at chem.vu.nl - http://www.chem.vu.nl/afdelingen/FAR|
|  /  \  |-----------------------------------------------------------|
| (    ) | Dept. of Biophysical Chemistry - University of Groningen  |
|  \__/  | Nijenborgh 4 - 9747 AG Groningen - The Netherlands        |
|   __   | Tel +31 50 363 4327 - Fax +31 50 363 4800                 |
|  /  \  | K.A.Feenstra at chem.rug.nl - http://md.chem.rug.nl/~anton   |
| (    ) |-----------------------------------------------------------|
|  \__/  | "If You See Me Getting High, Knock Me Down"               |
|        | (Red Hot Chili Peppers)                                   |

More information about the gromacs.org_gmx-users mailing list