[gmx-users] Question about omega torsion

[Erik Lindahl]

Hi,

On May 12, 2004, at 3:36 PM, Bert de Groot wrote:

>
> --Hua Wong wrote:
>> After some time fiddling with the parameters of the new minimisation 
>> engines (both CG or SD), I think I got them pretty much in hand now. 
>> It works pretty well.
>> However, I still find omega angle torsion deviation (using procheck), 
>> although lesser than with NAMD for example. Is there any other 
>> parameters I can refine in order to decrease (or make vanish) the 
>> number of peptide omega angles deviation?
>
>
> this is a known "feature" of most MD protein force fields. What you 
> can do is to
> increase the force constant for the omega torsion angle. For 
> minimisation you can do so
> without problems, but I wouldn't recommend it for MD, since it might 
> affect the dynamics
> in undesired ways.
>

The core of the problem is most likely that there are global errors in 
your structure, i.e. the sidechain can't adopt the minimum of the 
torsion without atoms colliding.

As Bert says though, don't change it, even if you are just interested 
in a single structure - you will just destroy your structure to make 
procheck happy (note that procheck would accept an unfolded peptide as 
OK, as long as the local interactions are in their respective minima).

Cheers,

Erik