Re: [Apbs-users] question about Coulomb energy and PME

[Lei Z. <lz...@co...>]

Dear Prof. Baker and apbs-user,

Thanks for the reply. I have another two questions related to the 
previous one:

1, I amm wondering whether APBS could be set with periodic 
boundary condition, if I want to use 
continuum electrostatics treatment to verify PME of explicit MD (like 
Huenberger's paper)?

2, how to modify the APBS input file so that it can be used as a poission 
solver (not PB solver) to calculate coulomb energy (like the Coulomb 
program)?

Thanks again for the help.

Lei Zhou



On Mon, 12 Sep 2005, Nathan Baker wrote:

> Hello --
>
> There are several differences between PME and straight Coulombic calculations 
> which could give rise to this constant offset.  The APBS calculation (using 
> the `coulomb` program) uses Coulomb's law with no cut-off and therefore 
> should give the most accurate calculation for this test case.
>
> I think your comment about the zero-wavelength PME correction for the 
> non-neutral system is probably a good guess.  This correction assigns a 
> uniform charge background to make the entire system neutral; this could lead 
> to the energetic offset you're seeing.  However, this should be easy to 
> check:  add a few counterions to your calculations and see if the results 
> agree with the Coulomb law prediction.
>
> Good luck,
>
> Nathan
>
> On 9/10/2005 9:42 PM, lz...@co... wrote:
>> Dear APBS-users,
>> 
>> Recently I have been working on a protein-ligand system and tring to
>> determine the coulomb interaction energy between the protein and
>> ligand. I got some strange difference comparing the energy
>> calculated by Coulomb (with APBS/tools/manip) and by PME in
>> gromacs.
>> 
>> The protein-ligand system was simulated with Gromacs package, with
>> explict water, PME for electrostatics and 1.0 nm cutoff. Because
>> gromacs energy only give the total PME (Coul-LR) for the system, I
>> have to turn off the charges for part of the system, rerun the
>> trajectory and then get the Coul-LR (PME energy) of protein or
>> ligand or protein-ligand. After that by summing Coul-SR and Coul-LR
>> of protein-ligand, I can get the coulomb energy between protein and
>> ligand.
>> 
>> To double check whether the results is valid, I used the Coulomb from APBS 
>> package (coulomb) to calculate the point to point coulomb
>> energy between protein and ligand (complex - protein - ligand),
>> using the gromacs partial charges. The pqr file was generated
>> automatically
>> by editconf from gromacs (with -mead option on).
>> 
>> The results from this two methods are very consistant from frame to
>> frame for the whole trajectory, except there is shift in baseline
>> about -1900 KJ/mol. In another words, the difference between those
>> two methods is a constant for the whole trajectory and that is
>> determinded by some intrinsic properties of the protein.
>> 
>> I tried to search the gromacs user list and just could not find a
>> clue where the difference come from. Is it related to the
>> requirement of PME for the total charge of the system to be zero?
>> Since I had to turn off the charges of other energy groups to get
>> the Coul-LR (PME energy) of a specific group, the net charges of
>> the system is not zero when I rerun the trajectory.
>> 
>> Or the PME result is more reliable in terms of treating long term
>> energies?
>> 
>> I am wondering whether you could give me some suggestions to solve
>> this puzzle. Thank you.
>> 
>> Sincerely,
>> 
>> Lei Zhou
>> _______________________________________________
>> apbs-users mailing list
>> apb...@ch...
>> http://cholla.wustl.edu/mailman/listinfo/apbs-users
>
> -- 
> Assistant Professor, Dept. of Biochemistry and Molecular Biophysics
> Center for Computational Biology, Washington University in St. Louis
> Web: http://cholla.wustl.edu/
>