From: Brian Radak (brian.radak.accts_at_gmail.com)
Date: Tue Jun 21 2016 - 07:36:06 CDT
On 06/20/2016 05:09 PM, Yue Qian wrote:
> Dear NAMD users,
>
> I am currently using NAMD to carry out free energy of binding
> calculations of a protein-ligand system. I have encountered several
> problems regrading the long-range LJ corrections. I have googled for
> certain issues but couldn't find relevant, informative discussions.
>
> This is what I found on the manual:
> LJcorrection: Description: Apply an analytical correction to the
> reported vdW energy and virial that is equal to the amount lost due to
> switching and cutoff of the LJ potential. The correction will use the
> average of vdW parameters for all particles in the system and assume a
> constant, homogeneous distribution of particles beyond the switching
> distance. See [60] for details (the equations used in the NAMD
> implementation are slightly different due to the use of a different
> switching function). Periodic boundary conditions are required to make
> use of tail corrections. LJcorrection as implemented is inconsistent
> with vdwForceSwitching.
>
> Here are my questions:
> 1. To turn on the long-range LJ correction, is "LJcorrection yes"
> is the only thing I need to do? (From the manual, I noticed that
> "LJcorrection as implemented is inconsistent with vdwForceSwitching."
> vdwForceSwitching was kept off.) I have tested this protocol in
> unbound state calculation with only the solute molecule and TIP3 water
> molecules with LJcorrection on and off. However, I did not observe
> significant difference, which was expected to be around 0.5 ~ 0.6
> kcal/mol. But with "Ewald off", the calculation result was
> ~0.8kcal/mol more positive. This patten applies to different cutoffs,
> from 8 Ang to 12 Ang.
>
Correct, "LJCorrection yes" or "LJCorrection on" will apply the
correction. As of 2.11 there actually is now a vdwForceSwitching
compatible correction, I just haven't updated the documentation yet
(sorry!). If a warning message still states this incompatibility, then
it is not present, because I also removed that message in the update.
The new code also improves consistency in conjunction with alchemical
endpoints, but might not work as expected with the (also mostly
undocumented) WCA FEP code. Standard FEP and TI should work more or less
ok now.
In general, such a correction will not significantly change free
energies unless the solute rather very large. Even then I might only
expect sub kcal/mol changes.
The main changes I expected when I did all of this was to get different
densities during an NpT equilibration. In general the correction
increases density by ~5%.
> 2. According to the manual, the LJcorrction option assumes a
> homogeneous, isotropic environment which only applies to unbound state
> calculation. To be able to apply to bound state calculation,
> heterogenous, nonisotropic environment needs to be considered. The
> cited paper (J. Phys. Chem. B, Vol. 111, No. 45, 2007) covered both
> scenarios. But GROMACS was used in their work. I was wondering if
> there's anyway to incorporate such methods in NAMD.
The main options here are to apply a Lennard-Jones Ewald approach (not
available in NAMD) or else to use the perturbation-like method by
re-evaluating the energy with a larger cutoff. The latter is doable but
has never been automated in NAMD to my knowledge (I suppose we could
give this a try if there is interest).
>
> Any clue is appreciated. Thank you very much.
>
> Best,
> Yue
>
> --
> Yue Qian
>
-- Brian Radak Postdoctoral Scholar Gordon Center for Integrative Science, W323A Department of Biochemistry & Molecular Biology University of Chicago 929 E. 57th St. Chicago, IL 60637-1454 Tel: 773/834-2812 email: radak_at_uchicago.edu
This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:33 CST