[gmx-users] Simulation

[Sergio Manzetti]

Didn't get an answer, so I try again:


Hi all, in the past I have had many experiences with MD and the negative 
effects on protein structures and models. In this regard, I keep certain 
residues freezed for a better simulation outcome, which highly reduces 
"negative divergence" from the start model. Believe me, but the start model 
(if its from protein modelling) can be the best structural candidate all 
the way to 20% homology and when this is simulated it all becomes worst.
In this regard I am trying various approaches to keep the reisudes from 
fluctuating in an exaggerated manner. One thing I thought about was to do 
simulations at 100K for a loong time, rather than 300K for a shorter 
period. Basically I am looking for smoother transitions to energy minimums 
because I think that the possibility to get stuck in a misfolded candidate 
is much higher at high temperatures given the higher number of candidates 
explored. In other words, I am looking for the "descense to a good minimum" 
from the initial model, rather than exhaustive search for a energy minimum.
Are there other ways I can do such an approach except lowering the temperature?
Sergio