[gmx-users] Simulated Annealing General Procedure

[Christian Blau]

Hello Neena,


The outline you give appears reasonable for exploring peptide conformations. It is very hard to judge appropriateness 
though without knowing what you are trying to learn from your system.

The unspecified position restraints might defeat the purpose of the simulated annealing if you keep too much of your 
system fixed. It's good to prevent chirality changes however, and back-bone position restraints are the right choice if 
you are interested in side-chain conformation changes.


If you are not tied to simulated annealing, but just want to explore your peptide conformations, the following methods 
are also useful:

  - if you know the reaction coordinate that your are interested in, e.g., a distance between residues, an angle or 
dihedral, then I suggest you try the "accelerated weighted histogram" (AWH) module. It may benefit your sampling a lot, 
because it ensures in a near optimal way that your system visits all conformations along a given reaction coordinate, 
e..g, all C-alpha - C-alpha distances between two residues in a given range

  - if you don't know the reaction coordinate, then Temperature-replica exchange for small peptide systems or flooding 
with essential dynamics sampling might be useful.


Best,

Christian

On 10/29/19 5:08 PM, Neena Susan Eappen wrote:
> Hello gromacs users,
>
> My goal is to do a better conformational sampling of my peptides using simulated annealing. Is my following procedure correct?
>
> 1- Energy minimization
> 2- NVT equilibration with position restraints say at 600 K for 100 ps
> 3- Simulated annealing with position restraints to cool down from 600 K to 300 K for 152 ps, with no velocity generation
> 4- Production simulation at 300 K for 10 ns without position restraints, with no velocity generation
> 5- Another step of energy minimization
> 6- Then repeat steps 2-6 multiple times with initially minimized structure from step 1
>
> Many thanks,
> Neena