[gmx-users] Implicit solvation: nonpolar term?

[Justin Lemkul]

On 12/19/13 7:33 PM, Robert Darkins wrote:
>
> On 19/12/13 23:31, Justin Lemkul wrote:
>>
>>
>> On 12/19/13 6:14 PM, rdwducl wrote:
>>> Thanks for the reply Justin but I'm confused. Why must the atoms be bonded?
>>>
>>> Ultimately I plan to model micellisation whereby surfactants, which only
>>> interact with each other via non-bonded terms, are driven to aggregate by
>>> this nonpolar term...
>>>
>>
>> I'm making a somewhat educated guess based on what you're observing. A
>> trivially simple test is to run two calculations with two particles in a box -
>> one in which a bond is defined between them, and one in which there is no
>> bond, but the atoms are at the same distance.  If the results are different,
>> my suspicion is confirmed (that the topology controls whether or not two atoms
>> impact one another's surface accessibility).
>>
>> Give that a try and see what happens.
>>
>> -Justin
>>
>
> You appear to be correct.
>
> In the md.log file /without/ any bonds:
>
>   Computing:                               M-Number         M-Flops % Flops
> -----------------------------------------------------------------------------
>   NB VdW [V&F]                             0.003682 0.004    49.2
>   NS-Pairs                                 0.000056 0.001    15.7
>   CG-CoM                                   0.000014 0.000     0.6
>   Virial                                   0.000104 0.002    25.0
>   Stop-CM                                  0.000014 0.000     1.9
>   Calc-Ekin                                0.000021 0.001     7.6
> -----------------------------------------------------------------------------
>   Total                                                       0.007 100.0
> -----------------------------------------------------------------------------
>
> whereas when a bond is added:
>
>   Computing:                               M-Number         M-Flops % Flops
> -----------------------------------------------------------------------------
>   Born radii (HCT/OBC)                     0.000244 0.045    96.5
>   Born force chain rule                    0.000024 0.000     0.8
>   NS-Pairs                                 0.000001 0.000     0.0
>   CG-CoM                                   0.000002 0.000     0.0
>   Virial                                   0.000047 0.001     1.8
>   Stop-CM                                  0.000004 0.000     0.1
>   Calc-Ekin                                0.000004 0.000     0.2
>   Lincs                                    0.000003 0.000     0.4
>   Constraint-V                             0.000004 0.000     0.1
>   Constraint-Vir                           0.000001 0.000     0.1
> -----------------------------------------------------------------------------
>   Total                                                       0.046 100.0
> -----------------------------------------------------------------------------
>
> So an atom will only 'see' another atom in computing the nonpolar term if it is
> bonded to it? That doesn't make sense in my mind. What am I missing?
>

Having not had the time to go into the code to see how all of this is working, 
the only sort of hand-waving explanation I can offer is this.  The Born radius 
characterizes how buried the atom is, so if two particles are bonded, they 
impact the accessibility of each other since a chemical bond exists between 
them.  In real life, that would represent electron density, and not empty space 
that could be occupied by something else.  You could argue that at some very 
small distance, even if not bonded, the atoms will contact one another and 
influence solvent accessibility; perhaps if this is not the case for very small 
distances (< 0.1 nm) then there is a bug to be investigated with respect to 
surface generation.  Of course, the likelihood of anything coming that close via 
nonbonded interactions is pretty small, though that depends on the repulsive 
potential between the two atoms.

-Justin

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

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

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