Juan,
I have also tried to use the weighting factors in RESP fitting, but have
not carefully tested them in any way other than setting different
conformers to equal (or pretty similar) weights. In principle, one would
use the relative energies to determine Boltzmann factors which could become
your weights. My belief is that the weighting factors you put into the RESP
input file get normalized by RESP, so you can make the values whatever you
like as long as their relative sizes are appropriate.
In reality I think that determining these weighting factors is pretty
difficult. One should really have free energies (rather than just
energies/enthalpies) including an environment that represents the average
that your residue will see in your simulations. In the standard protocol
one uses the cheap HF/6-31G(d) in vacuum method for determining ESPs
because the errors in this method lead to charges that are polarized in a
way that is approximately correct for water solvation. If you will be using
your residue with Cornell et al. type residues (e.g. ff94, ff99, ff99SB,
ff14SB) then it is critical that you use this same approximate method
because it is more important that your charges be balanced relative to
those FFs than that your charges represent 'reality' for that residue.
Anyway, you can certainly do some quick testing using different weights to
see how much impact they have on the final charges. I would personally use
just those conformers with low energies (as you mentioned).
I never use R.E.D. myself, so I can't offer any help with that.
Good luck!
Cheers,
Brent
On Tue, Jun 6, 2017 at 7:05 AM, Eiros Zamora, Juan <
j.eiros-zamora14.imperial.ac.uk> wrote:
> Dear Amber experts,
>
> I am on the process of deriving point charges for a small molecule of 51
> atoms. From is SMILES string, I’ve generated several conformers using RDKit
> (http://www.rdkit.org/). My objective is to do multi-conformational RESP
> fitting. I’ve seen on the Advanced Tutorial1 http://ambermd.org/tutorials/
> advanced/tutorial1/section1.htm that the recommended way to do this for
> reproducibility is using the R.E.D server. However I decided to manually do
> it myself too to better understand the process.
>
> Starting from 9 conformers, I’ve noticed that some of them are differ in
> energies (between 3 to 10 kcal/mol to the lowest energy conformer, Gaussian
> SPE calculation and M06/cc-pvdz). Should I only use for resp the conformers
> that are around < 2kcal/mol to the lowest energy conformer? I think this
> can be also done by tweaking with the weighting factor for the
> multi-molecule fit. But in the tutorial, it just states that is set to 1.0
> for both conformers since the two have to be weighted equally. I’ve been
> looking in the manual and haven’t seen much being discussed about this
> weighting factor. Does anyone know how to use it to better represent the
> energy difference between the conformers?
>
> Additionally, and this is related to the R.E.D server: is it possible to
> modify these weighting factors according to the energy difference between
> the conformers (after optimization, presumably)? I’ve set up a job with 9
> conformers of the molecule in a single PDB file, but upon inspection of the
> job results, I’ve observed that these weighting factors are all set to 1.0.
> I’ve also noticed that the input file for resp in the R.E.D server doesn’t
> set the nmol keyword to however many conformers there are – instead, on top
> of every molecule section something like this is set: MOLECULE-1 nconf=1
> norient=1 nmep=1/18 (the nconf, norient and nmep values change). Can anyone
> comment or point me in the right direction to better understand the input
> files for resp? I’m also a bit confused by the fact that there are 18
> entries in this resp input file when I only loaded 9 conformers.
>
> Thanks for any suggestions!
>
> Juan Eiros
>
> [Imperial College London] <http://www.icb-cdt.co.uk/>
>
>
> Juan Eiros Zamora / PhD Student
> je714.ic.ac.uk<mailto:je714.ic.ac.uk> / +44 (0) 7783018326
>
> Imperial College London
> Imperial College Road SW7 2AZ | UK
> http://www.icb-cdt.co.uk/
>
>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
_______________________________________________
Brent P. Krueger.....................phone: 616 395 7629
Professor.................................fax: 616 395 7118
Hope College..........................Schaap Hall 2120
Department of Chemistry
Holland, MI 49423
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jun 06 2017 - 05:00:02 PDT
