[gmx-users] Does gmx covar/gmx anaeig give <dS> or T<dS> for ligand binding?

[David van der Spoel]

On 21/06/16 11:26, Billy Williams-Noonan wrote:
> Hi Gromacs Users,
>
>   I have used gmx covar and gmx anaeig to generate three ensemble average
> entropies over 100ns: first for a ligand in solution (<S(L)>), second for a
> protein in solution (<S(P)>) and third for their respective complex in
> solution (<S(P.L)>).
>
>    My understanding is that the change in entropy upon binding is given by:
>
> <dS> = <S(P.L)> - <S(P)> - <S(L)>   -----------(1)
>
>    Using gmx covar/gmx anaeig I got Quasi-Harmonic entropy estimates of:
>
> <S(P.L)> = 128,886 J/mol/K
>
> <S(P)> = 153,548 J/mol/K
>
> <S(L)> = 4137 J/mol/K
>
>    As stated, these values were generated using gmx covar/anaeig by
> selecting for the relevant biomolecule in each ensemble and ignoring the
> effect of solvent movement.
The unit printed by the program is J/mol K, which is the normal unit for 
entropy in all handbooks. You can not prove or disprove the correctness 
of the code by an example, you will have to check the code yourself if 
you doubt it.

Looking at your numbers, they are huge and the difference is huge too.
You should probably make sure first that all you simulations are in 
equilibrium. A ligand entropy och 4137 I would expect for an organic 
molecules with close to 100 carbon atoms.


>
>    By subbing the above-described values into (1), I got about -28 kJ/mol/K
> for <dS>, which is the right answer if the units are actually kJ/mol, and
> not kJ/mol/K.  Strangely, upon multiplying by T, I got a value of -8640
> kJ/mol, which is quite obviously wrong.
>
>    So does (1) yield a value for <S> or T<dS> ?  Is anyone able to explain
> this to me?
>
>    Kind regards,
>
> Billy
>
>


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se