Hi,
If you actually look at the computational cost of explicit solvent
simulations with PME they can be comparable to GB simulations
(particularly for larger solutes), especially if you're using GPUs.
The real "savings" you get from implicit solvent simulations is in
sampling since the water is just a field that can relax instantly to
changes in solute conformation - you don't have to "push" water out of
the way to move around. Implicit solvent can be fine for some systems,
but can also give poor results for certain molecular "shapes" (because
they don't handle anisotropy well), or for highly charged systems
(boundary effects), or for system where structural waters may be
important, etc. Also, many implicit solvent models have a poor
description of the nonpolar component of the solvation energy (surface
area only based terms are not where you want to be), which can matter
if you expect your solute to change shape (e.g. during
folding/unfolding).
Just my $0.02...
-Dan
On Wed, Jun 7, 2017 at 6:00 AM, <diego.soler.uam.es> wrote:
> Hi Dading,
>
> It depends on what you want to do. Implicit solvent models use a lot
> of approximations, such as the linearization of the Poisson-Boltzman
> equation (Generalized Born model), while explicit solvent can be made
> as realistic as you want.
> Implicit solvents are good to describe energy transfer in solvation
> proccess, for example, and since they are quite efficient and fast, it
> can be convenient to use them for a lot of problems.
>
> But an implicit solvent has several limitations (just the very nature
> and justification of the models makes it obvious) and explicit
> solvent, although clearly more expensive, will almost always be more
> accurate. But, of course, even if it is more accurate, sometimes you
> won't care about the extra degree of accuracy that an explicit solvent
> can provide and you should prefer the efficiency of an implicit model.
> As usual, this is all problem-dependent. If you share the particular
> problem you are thinking of maybe someone can give you further insight.
>
> Cheers,
>
> DiegoSP
>
>
> Quoting Dd H <ddhecnu.gmail.com>:
>
>> Thank you for your reply, Bill. What is the meaning of "probably"? Is there
>> a case that implicit solvent is more accurate than explicit solvent?
>>
>> Best regards,
>> Dading Huang
>>
>>
>> On Wed, Jun 7, 2017 at 4:09 PM, Bill Ross <ross.cgl.ucsf.edu> wrote:
>>
>>> I believe people use explicit solvent when they can afford to, because
>>> it is more realistic and probably more accurate.
>>>
>>> Bill
>>>
>>>
>>> On 6/7/17 1:06 AM, Dd H wrote:
>>> > Hi Amber users,
>>> >
>>> > It seems like explicit solvent is more commonly used for MD simulations
>>> > now. Can you tell me the reason why people choose explicit solvent rather
>>> > than implicit solvent? Does explicit solvent gives more accurate sampling
>>> > than implicit solvent?
>>> >
>>> > Best regards,
>>> > Dading Huang
>>> > _______________________________________________
>>> > AMBER mailing list
>>> > AMBER.ambermd.org
>>> > http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 07 2017 - 07:00:04 PDT
