[gmx-users] Re: pKa calculations using gromacs

[Markus O Kaukonen]

Thank's Shameer for the nice reference.

However, after reading the article two issues are still unclear to me

1) In
>Simonson, T., J. Carlsson, D.A. Case. 2004. Proton binding to proteins:
>pK(a) calculations with explicit and implicit solvent models. J Am Chem
>Soc 126(13):4167-80.
they do thermodynamic integration
from system A (protonated Asp) to system B (ionised Asp).
They use periodic boundary conditions and explicit water molecules to fill
the simulation box around protein (and within protein).
They add counter ions to neutralise the box in state A.
It is not said in the article if they add one more conter ion to state B.

Question:
Is this really possible?
In one system (say A) they have
total charge of 0 in the system (after adding counter ions)
, but in B they have either total charge -1
(or they have added one more Na+ in the system which is not said in the text).
Does not one run into troubles with electrostatistics when the simulation
cell is not neutral (authors have used Charmm and Amber).

On the other hand, if one would add one more counter ion to state B
(to make its charge 0) this would add up uncorrect energy terms to state
B.

Well, lets stop here for the moment .

Terveisin Markus

>Although linear response theories using MD free energies have now been
>used for quite a while to predict amino-acid pka's, it definitely is NOT a
>completely reliable method. You can find some of the pros and cons of
>this method discussed in a recently published journal article:

>Simonson, T., J. Carlsson, D.A. Case. 2004. Proton binding to proteins:
>pK(a) calculations with explicit and implicit solvent models. J Am Chem
>Soc 126(13):4167-80.

>You would find that different force-fields produce different directions
>of pKa shifts!
>But is it really just a force-field related issue? I think not.

>cheers - Sameer


>Dear All,
>
>Did some literature search and could not find any hit with
>'pKa AND Gromacs'. Maybe there is a reason for this?
>
>I wonder does somebody know about calculting of gibbs free energy of
>AH(p) + H2O(aq) -> A-(p) + H3O+(aq)
>A   : titrable residue in protein,(p) : in protein, (aq): in bulk water
>using Gromacs?
>
>Based on this deltaG one gets pKa estimate of the residue
>assuming a given protonation state of the rest of the protein
>(see for example Schuurmann J.Phys.ChemA v102, pp6706-6712-y1998,
>                  Sham J.Phys.ChemB v101, pp4458-4472-y1997 ).
>
>So to the question:
>I guess in the actual calculation one must use periodic boundary
>condition and this requires the simulation box to be neutral.
>
>The only way I was able to fullfill this requirement is simply
>to make a big water box containing the protein and putting Na+
>atoms to water part to make simulation box neutral.
>
>In AH(p) + H2O(aq) state the H2O(aq) is 'far away' from the protein in
water
>(state X).
>In A-(p) + H3O+(aq) state the H3O+(aq) is 'far away' from the protein in
water
>(state Y)
>
>Then thermodynamic integration between states X and Y to get deltaG
>(and pKa). (See excellent examples in
>http://md.chem.rug.nl/education/Free-Energy_Course/)
>
>Is there some serious flaw in this procedure because I could not find
this
>in the literature?  (Like pH of the water in A-(p) + H3O+(aq) would be
-3,
>no force field parameters for H3O+...). Partial charges for A- and AH one
>gets from QM.
>
>Terveisin Markus



-- 
-----------www= http://www.iki.fi/markus.kaukonen---------------
Markus.Kaukonen at helsinki.fi   \ HOME->   Verajakallionkatu 6 a 5
Helsinki Bioenergetics Group   \_______  02600 Espoo, Finland
Institute of Biotechnology             \ tel:+358-(0)9-5127122/h
Biocenter 3, P.O.BOX 65 (Viikinkaari 1) \____
FIN-00014 University of Helsinki              \______ H+ ______
tel: +358-(0)9-19159752/o, fax: +358-(0)9-19159920             \
--- Rikos ei kannata, eika maatalous Suomessa. (Paimio 1998) ---