[gmx-users] Re: Counterions: influence on protein

[Mark Abraham]

Maxim Fedorov wrote:
> Dear Mark, 
> 

>>Charge separation is expensive energetically. Even though the 
>>unfolding 
>>may be charge-driven, it simply isn't going to be true that there 
>>are no 
>>counter-ions around in an experiment. Thus to have a realistic 
>>simulation you need to have some counter-ions (preferably at a 
>>realistic 
>>ionic strength)... but you say that "some counter-ions" stabilize 
>>the 
>>initial conformation. If you were adding charges to neutralize the 
>>system, you could try adding fewer counter-ions as a compromise.
> 
> But in this case I would have non-zero net charge of the box - 
> how can I use PME then?

It's totally transparent. The basic unit of charge in Ewald methods can 
be thought of as a point charge and a so-called homogenous neutralizing 
background charge density - every charge effectively has this, so the 
total charge of the "natural" system is immaterial. Whether that's a 
good model physics is a different matter... Check out some refs - one 
paper I have cites G. Hummer, L. R. Pratt, and A. E. Garcia, J. Phys. 
Chem. 100, 1206
1996, B. R. A. Nijboer and T. W. Ruijgrok, J. Stat. Phys. 53, 361 1988,
B. Cichocki, B. U. Felderhof, and K. Hinsen, Phys. Rev. A 39, 5350
1989, G. Hummer and D. M. Soumpasis, J. Chem. Phys. 98, 581 1993 in 
reference to this.

>>Whatever you do to change the system is going to be a non-physical 
>>system. You say the physically realistic simulation doesn't follow 
>>experiment. Perhaps you should have a look at your simulation 
>>protocol, 
>>particularly the electrostatics treatment, to see if the fault 
>>might be 
>>there, rather than assume it's only the charge-screening effect. 
>>MM 
>>force fields are a model of reality, often applied in MD using 
>>further 
>>approximations to real physics, so there are multiple sources of 
>>problems.
> 
> Indeed. My general impression now is that SPC + PME is not a good 
> combination 
> for this task. The SPC water is too rigid and as was shown by 
> Dzubiella and Hansen in 
> a couple of theirs recent papers it tends to form some ice-like 
> structures around charges.
> The PME itself can also produce some overstabilisation - see Roberto 
> D. Lins and Ursula Röthlisberger J. Chem. Theory Comput.; 2006 (many 
> thanks to an unknown friend who forwarded my question to Roberto!).
> So, I guess, I should play a bit with different water models. I am a 
> bit afraid to change the PME by cut-off though. This poly-L-lysine 
> business is only part of 
> our tasks and for other systems the PME works better than cut-off. 

Indeed.

Mark