[gmx-users] Re: Justin umbrella sampling tutorial......

Thomas Schlesier schlesi at uni-mainz.de
Fri May 25 17:16:40 CEST 2012

I think all the answers to your question are in the tutorial. Probably 
read first the  lysozyme tutorial and then the umbrella tutorial again.

But here is a more general answer:

Normally you have two types of simulations:
preperation (which is also equilibration)
and the need of restraints depends on what you want to achive.

So if you want to calculate some equilibrium property of a protein in 
water you do first a preperation simulation to equilibrate the system 
(NVE, NVT or NPT - depending for what ensemble you want the property).
Normally during this you put position restraints to the protein 
backbone, so that the protein structure does not gets disturbed during 
the part where you equilibrate the water. but if your protein is fairly 
stable / rigid, you don't need these restraints.
Then you do the production simulation. Normally without restraints, 
since you want to know the property for a protein in equilibrium. But 
you could also do this with the same position restraints, in the extrem 
case, when you freeze the protein, you would get the property for a 
special protein geometry (the one to which ou restrain the coordinates).
-> So it really depends on what you want to do!!!

For the umbrella tutorial:
As far as i remember, Justin used the restrains for one of the chains 
(B), so to conserve the structure of said chain, because when one pull 
chain A away this would have an influence to the structure of chain B 
(they interact with each other).
BUT: one could also dismiss these restrains during the umbrella 
simulations, since it is expected the structure of chain B depends on 
the distance to chain A (since they interact).
You could also restrain only that part of chain B, which does not 
interact directly with chain A, to conserve a part of the structure of B 
(so you would be in the middle between the two extreme cases above).

In all three cases you get a different answer (like above for the 
equilibrium property). One could argue which is the better option, but i 
think all three have their right to exist, since the answer to this also 
depends on the initial question. Best would be to do all three cases, 
but mostly one does not have so much time. So one must stick to one.

Which leads to:
First think about your question.
Then think about how to answer it (think often there will be different 
possibilities / methods).
Then decide what to do...
Since the answer you get (may) depend on the method you use, you should 
justify your decission.

Hope this helps

>> Thank you Justin for these correct explanation
>> >  Its really clear my lot of queries..
>> >
>> >       For the tutorial, NPT is conducted with restraints on all protein heavy
>> >       atoms.  The production runs are conducted by restraining only one chain for
>> >       practical reasons.
>> >
>> >       These is my question ;
>> >
>> >  If we are doing NPT with restraints on all protein atom and production run by
>> >  conducted by restraining only one chain for protein...
>> >
>> >  means NPT and productin run mdp files should be different , Where these
>> >  information in mdp files???
>> >  It my request to you please tell me why these mdp files are almost same in
>> >  parameter ..
>> >
>> >  (Where you mentioned in mdp for npt to do restraints on all proteins heavy atoms
>> >  and for production md restraining only
>> >  one chain ..)
>> >
> I don't understand what more I can say beyond what I already have.
> -Justin

More information about the gromacs.org_gmx-users mailing list