[gmx-users] free energy and ions
spoel at xray.bmc.uu.se
Wed Nov 27 23:57:05 CET 2002
On Wed, 2002-11-27 at 20:04, Alan Wilter Sousa da Silva wrote:
> Hi List,
> I'm trying to calculate binding free energy of a complex
> substrate/protein using linear approximation (Hansson & Aqvist, Protein
> Eng. 1995, 8, 1137-1144).
> Both protein and substrate are charged, so I used ions and PME.
> However, observing dynamics (1 ns) I can see that for the complex ions
> keep around substrate during whole simulation, while MD of the only
> substrate in water a ion start close from it but after 100 ps it goes
> Using g_energy I could note a great difference when analyzing the
> first 100 ps (for both simulation) to the last 100ps, all due to the
> interaction between ions and substrate.
> In other words: Coul-SR:Substrate-CL- energy for complex
> oscillates with time (never zero) while Coul-SR:Substrate-CL- energy for
> substrate goes to zero in only 200 ps and there it remains.
> I emphasized Coul-SR:Substrato-CL- because it change completely my
> energy binding estimations.
> With substrates not charged I got really good results (compared to
> experimental values).
> How can I keep a coherence for analysis?
> How bad would it be if I discard ions and rerun both simulation?
> Should I run more 1 or 2 ns and so get a more reliable statistic?
There is no easy answer for this question... Ideally you'd want to have
real equilibrium distributions of ions. However you apparently don't see
that on the ns timescale. If you leave out the ions you also have a
problem. You didn't state the charges on substrate and ligand, but if
they are low you maybe could get away with a cut-off and maybe an
analytical correction to the energy. Check the literature for Born
Dr. David van der Spoel, Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576, 75123 Uppsala, Sweden
phone: 46 18 471 4205 fax: 46 18 511 755
spoel at xray.bmc.uu.se spoel at gromacs.org http://zorn.bmc.uu.se/~spoel
More information about the gromacs.org_gmx-users