[gmx-users] General query regarding MD simulation.

[Justin Lemkul]

On 2/15/13 4:00 PM, Abhishek Acharya wrote:
>
>
>>
>> On 2/15/13 1:29 PM, Abhishek Acharya wrote:
>>> Dear GROMACS Users.
>>> Just out of curiosity, i would like to pose a general question here ( i
>>> didn't have an idea of  any other suitable forum ). My protein active
>>> site
>>> has a GDP which is coordinated to a Mg ion. The Mg ion itself
>>> coordinates
>>> two water molecules and is held in position via non-bonded interactions
>>> from two active site residues. For such a system, I could do two things:
>>>
>>> 1. Do the charge calculation for GDP only and assume the charges of
>>> other
>>> active site constituents to be taken from the FF parameter library.
>>>
>>> 2. As suggested by a a person I know, I can do the charge calculation of
>>> the whole system including the Mg ion, water molecules and the residues.
>>> The explanation was that since GDP is in coordination to Mg ion, the
>>> effective charges would be different than on a GDP alone.
>>>
>>> Can anyone explain which one of the above is a correct approach and why
>>> ?
>>> I somehow was not convinced by my co-workers explanation simply
>>> considering the fact that for each of the amino acids in a protein the
>>> charges and parameters are taken from the FF library. Going by the given
>>> explanation, one should then resort to a charge calculation for the
>>> whole
>>> protein system.
>>>
>>
>> In the context of normal MM force fields with fixed charges, option (1) is
>> what
>> would generally be used.  In determining what is more representative of an
>> actual biological setting, option (2) is more rigorously correct.  Force
>> fields
>> are usually parameterized in a portable way, such that every residue has
>> uniform
>> parameters independent of its local environment.  Thus polarization
>> effects are
>> treated in an average way, which may not be optimal.  Metal ions have
>> especially
>> polarizing effects on partial charges of nearby residues.  Even QM/MM
>> studies
>> that are 15 years old concluded that fixed charges for such systems are
>> inherently deficient.
>>
>> I guess the bottom line is you have to derive suitable parameters in a way
>> that
>> is compatible with the original force field.  If that means dealing with
>> the
>> ligand in isolation, so be it.  The comparison between the parameters
>> produced
>> by options (1) and (2) would be very interesting, though, and may
>> ultimately be
>> necessary in justifying why your proposed model worked (or didn't).
>>
> I think i would try both approaches.
> In regard to the derivation of ff compatible parameters, i see that for
> ff's like AMBER94 one resorts to ab-initio 6-31G* calculations and fitting
> to electrostatic potential surfaces. But for OPLSaa, if one goes through
> the literature, it says,  'the charges for OPLS are empirical and have
> been obtained largely from fitting to reproducible properties of organic
> liquids.The charges for functional groups are taken to be transferable
> between molecules and the use of neutral subunits makes the derivation of
> charges for large molecules straightforward.'
> I presume the first line means charges were derived from experimental
> data. But I didn't understand what the last line means. Kindly help!!

Empirical charge derivation, in a sense, means the authors fiddled with it until 
it worked ;)  In seriousness though, generally some preliminary QM calculations 
are done to obtain charge distribution, and then adjustments made to fit with 
known experimental data.  For OPLS, it's typically properties of liquids that 
serve as model compounds for the desired molecule.  The derivation is thus done 
in terms of functional groups, not necessarily whole molecules.  That's where 
the last line comes in - you can piece molecules together based on universal 
(supposedly), transferable groups whose parameters do not depend on the rest of 
the molecule.  For instance, GDP would not be derived directly, but its 
components would, and then the molecule would be pieced together from various 
functional groups (imidazole, amines, carbonyl compounds, etc).

-Justin

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

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

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