- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jason Swails <jason.swails.gmail.com>

Date: Sat, 7 Apr 2012 15:09:03 -0400

On Sat, Apr 7, 2012 at 2:57 PM, kirtana S <skirtana4.gmail.com> wrote:

*> Sorry for not using the right terminology . I have used explicit solvent
*

*> simulations.
*

*> I want to calculate the change in solvation free energy of polymer in
*

*> presence of ions.
*

*>
*

*> Is this only for implicit solvent models.
*

*>
*

The MM/GBSA methodology is pretty much limited to implicit solvent models.

The problem with using explicit solvent models with these types of

calculations is that convergence is nearly impossible. For

moderately-sized periodic boxes, the total energy is dominated by

solvent-solvent interactions, which means that any protein-solvent or

protein-protein interactions will be drowned out completely for a handful

of snapshots. The only way you can get meaningful statistics is if you use

so many snapshots that the noise contributed by solvent-solvent

interactions is averaged out.

Keep in mind that noise will drop off as 1/sqrt(N) where N is the number of

frames in your averaging -- therefore, for large noise (which this is), it

will require a _large_ number of snapshots to get your desired interactions.

Implicit solvent, on the other hand, models equilibrium solvation

properties -- i.e., it should average over the solvent degrees of freedom,

completely eliminating the noise caused by explicit solvent.

Note that implicit solvent models can incorporate generalized ionic effects

via the introduction of a debye screening parameter (for GB) and the

Boltzmann part of the PB equation which can be truncated after the first

order term in the salt concentration (linear PB) or carried out to the sinh

term (non-linear PB). The manual for MMPBSA.py explains how to set salt

concentrations (or ionic strengths) for MM/GBSA and MM/PBSA.

Of course this will not capture the effect of structurally important ions

-- this is a much more difficult problem.

HTH,

Jason

Date: Sat, 7 Apr 2012 15:09:03 -0400

On Sat, Apr 7, 2012 at 2:57 PM, kirtana S <skirtana4.gmail.com> wrote:

The MM/GBSA methodology is pretty much limited to implicit solvent models.

The problem with using explicit solvent models with these types of

calculations is that convergence is nearly impossible. For

moderately-sized periodic boxes, the total energy is dominated by

solvent-solvent interactions, which means that any protein-solvent or

protein-protein interactions will be drowned out completely for a handful

of snapshots. The only way you can get meaningful statistics is if you use

so many snapshots that the noise contributed by solvent-solvent

interactions is averaged out.

Keep in mind that noise will drop off as 1/sqrt(N) where N is the number of

frames in your averaging -- therefore, for large noise (which this is), it

will require a _large_ number of snapshots to get your desired interactions.

Implicit solvent, on the other hand, models equilibrium solvation

properties -- i.e., it should average over the solvent degrees of freedom,

completely eliminating the noise caused by explicit solvent.

Note that implicit solvent models can incorporate generalized ionic effects

via the introduction of a debye screening parameter (for GB) and the

Boltzmann part of the PB equation which can be truncated after the first

order term in the salt concentration (linear PB) or carried out to the sinh

term (non-linear PB). The manual for MMPBSA.py explains how to set salt

concentrations (or ionic strengths) for MM/GBSA and MM/PBSA.

Of course this will not capture the effect of structurally important ions

-- this is a much more difficult problem.

HTH,

Jason

-- Jason M. Swails Quantum Theory Project, University of Florida Ph.D. Candidate 352-392-4032 _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Sat Apr 07 2012 - 12:30:03 PDT

Custom Search