[gmx-users] Re: Simulation in the high temperature conditions

[Justin A. Lemkul]

James Starlight wrote:
> Mark,
> 
> 
> 
> 
> 
>     Assuming you're raising your temperature during equilibration and
>     then running at high temperature, then you don't want water moving
>     into the receptor interior during equilibration for the same reason
>     you didn't want water moving into the CCl4. And you're going to run
>     further equilibration after taking off all the restraints anyway,
>     right? And if water moves into the receptor interior, then it
>     probably does that under high-temperature equilibrium conditions...
> 
> 
> It is that I want to prevent movement of water into Ccl4 layer (that is 
> drive by the increased temperature and have not been present in the 
> natural conditions) but allow it to move into the receptor interiour ( 
> that is normally observed experimentally). So the posres application may 
> prevent water to move both into the receptor and Ccl4 so this aproach 
> could not be useful in my case, couldn't it ?
> 
>  

No, it wouldn't.  I don't understand why you believe some water absolutely can't 
leak into the CCl4 layer, at least transiently.  If you're getting massive 
mixing, then that's a problem for which I don't know a clear solution in this 
case.  Occasional leakage is not unexpected, as it even occurs in lipid 
bilayers.  True, water and CCl4 are not particularly miscible, but to say that 
they should remain 100% isolated in reality is not correct.

> 
> 
>     On point, the reference temperature has little to do with whether
>     phases diffuse into (or out of) each other, and lots to do with what
>     ensemble you might be sampling. The actual temperature controls the
>     rate of such diffusion, of course, but if the non-bonded
>     interactions allow for intermixing, then you'll get some degree of
>     that regardless of any other setting. You'd be well advised to check
>     that your CCl4-water boundary behaves acceptably before you invest
>     in the protein simulation...
> 
> 
> Specifically I've desided to rise temperature of my system to increase 
> conformation sampling of the protein surrounded of the interiour. For 
> this I've devided my system on several parts to couple to separate 
> thermostates. One part is the protein and CCl4 layer wich mimick the 
> membrane and another part is the sorrounded water ( SOL), internal water 
> ( wich may present in the protein interiour transitory but never in the 
> Ccl4 layer) and IONS ( wich are always in the SOL layer). In order that 
> increase the rate of conformation sampling I've gradually increased 
> ref_t of each t_group. In part this result in evaporation of water and 
> some of that water has moved in Ccl4 layer. So its intresting for me 
> might  I set ref_t for water_ions group as the specified value ( ref_t= 
> 310 k) and gradually increase only ref_t for the protein_ccl4 group in 
> the annealing manner. As the consequence I want to prevent evaporation 
> of the water but maintain of the overal temperature high of my system tp 
> increase sampling of the protein conformation ?
> 
> I inderstand that such aproach is very unphysical but aplication of the 
> posres and other tricks are from this theme but works good in some cases :)
> 
> 

This sounds really fishy to me.  Coupling different parts of a system to 
thermostats at different temperatures just begs for artifacts.  The use of 
thermostats for separate parts of the system, even at the same temperature, is 
questionable since you violate energy conservation.  Of course, for many cases, 
they're a necessity and we just go ahead with it ;)  If you've got two parts of 
a system in thermal disequilibrium, then you've got a very odd system with 
thermostats exchanging different amounts of energy with different portions of 
the system.  I don't know how to predict the effects of such a setup, but I 
don't recall having seen anything like this done before.

-Justin

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================