[gmx-users] gmx insert-molecules

[Justin Lemkul]

On 6/7/17 7:33 PM, Shi Li wrote:

>>> I am trying to avoid the long step of equilibrium as I have many systems corresponding to different concentrations. I was thinking if I replace a small number of molecule A with molecule B (the system A is very large and pre-equilibriumed) then I only need to apply a short time-step of NPT in order to let the system expand or shrink. Then I can use the new system to continue replacing A with B to generate a new concentration. Is this practical?
>>>
>>> The problem is when B is slightly larger than A, I can?t insert the same number of B into the system. Is there way to avoid the overlapping or force the molecule in?
>> You can reduce the vdW radius of atoms to *try* to force the molecules to fit,
>> but then all you've done is introduce bad clashes that have to be subjected to
>> minimization and re-equilibration.  So at that point, all you've done is build
>> your system in the most inefficient way possible.  By trying to avoid
>> equilibration, you've necessitated it :)
>>
>> Build the system the robust way - solute first, then solvent.  It's ultimately
>> less work and less prone to failure.
>>
>> -Justin
>>
>> -- 
> 
> 
> Thank you Justin.
> 
> In the way solute first then solvent, I will still need to fully re-equilbrium the system, is that right? My problem is that I have about a hundred of systems, each of them will have about half a million atoms. The full equilibrium will be too time consuming. Since each systems are only slightly different in concentration (by replacing molecule A with molecule B), I am wondering if there will be a easier way to do a quick/roughly equilibrium.

Again, please don't reply to the entire digest.

Neither approach avoids having re-equilibrate the system.  Your approach of 
inserting B into an existing box of A perturbs the system and requires a new 
equilibration.  This is especially true if you play tricks like messing with vdW 
radii to force B into small voids in A.  You need to minimize and equilibrate, 
because these are new systems.

My approach is less prone to failure and ultimately I would expect it to take 
less time overall because you will not have instances in which the insertion of 
B into A fails and requires you to revisit those systems, change seeds, hack vdW 
radii, and potentially deal with problematic minimizations.

-Justin

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

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

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