CHARMM Development Project
Posted By: Aravinda GFP chromophore parameters - 06/26/19 05:11 AM
I am planning to model a GFP protein. However, I could not find charmm parameters for the chromophore within charmm distribution. But, I found parameters derived by Reuter et al ( to be widely used by others who worked with GFPs. My question is, since these parameters were derived using C22, can I use those parameters and model the rest of the protein with C36 parameters?
Thank you,
Posted By: rmv Re: GFP chromophore parameters - 06/26/19 05:26 PM
There are a number of good reasons why the C22 force fields have been deprecated.

If the C22 chromophore parameters are widely used, perhaps someone has already used them with C36 parameters and published the result.

You can always try it, but if the chromophore will be a key part of the study, you should probably look into it more carefully.
Posted By: Aravinda Re: GFP chromophore parameters - 07/08/19 09:06 PM
Thank you very much for your thought on this. I went over most of the cited papers. But almost all of them (even ones published in 2018) used C27 forcefield along with these parameters. But as far as I know, C27 for proteins are the same as for C22. So, basically, they used C22..

Going over the chromophore (see below) and the following thread,
is it reasonable for me to take analogue parameters from C36 FF and build a whole new parameter/topology definition?.

! Hd2 Hb O2 H
! | | || |
! He2 Cd2 Cb C2 CT2--H
! \ / \\ / \\ / \ / \
! Ce2 Cg Ca N3 H
! || | \ /
! Cz Cd1 N2=C1 H
! / \ // \ \ /
! Oh- Ce1 Hd1 CT1-H
! | \
! He1 H

For Tyrosine analogue I can take parameters from Phenol derivatives. But, if I do this, what is the appropriate analogue for imidazole ring would be?

Attached picture chromophore.png
Posted By: rmv Re: GFP chromophore parameters - 07/08/19 09:34 PM
You could try the CgenFF server for that part (replaces ParamChem server).
Posted By: Aravinda Re: GFP chromophore parameters - 07/09/19 04:07 AM
I did try to use CGenFF to obtain imidazole ring associated parameters. But CGenFF website gave param penalty of 187.500 and charge penalty of 47.153 for analogue parameters.

(1) I can use FFTK within VMD or follow charmm guidelines to generate parameters and charges for the whole chromophore from scratch.
(2) Use analogue parameters from phenol derivatives for deprotonated tyrosine and, use FFTK or follow charmm guidelines to derive charges and parameters for the imidazole ring
(3) Use analogue parameters from phenol derivatives for deprotonated tyrosine and, all other parameters taking from already published work (
(4) Just simply use already published parameters for the whole chromophore along with C36m for the rest of the protein and see if chromophore within GFP matches experimental observations for validation.

If I am to chose from these four options, which one will you recommend?.
Also, I am wondering, what limits me using parameters derived for C22 to be used in C36? especially considering parameterization guidelines for novel residues are the same for C22 and C36 (please correct me if it is not)

p.s-sorry for the long request, I wish to know how exactly charmm developers look at this issue for me to learn from it. I am a bit reluctant to develop parameters for a system like this, especially when parameters from Reuter et al itself was never incorporated to charmm though it has been cited a lot.
Posted By: rmv Re: GFP chromophore parameters - 07/09/19 04:45 PM
Many of the L-J parameters and partial charges have changed since C22 releases, so there is some risk.

Of course, the more rigorous approach is to do some QM calculations, which is how the FFs were developed. CGenFF is derived from both the existing FFs, and some new QM calculations and validation simulations.

Using existing C36 parameters for fragments is encouraged, when possible, so the Tyr ring parameters could probably be used. The exocyclic C=C may pose the biggest challenge, in that it is not well represented in existing parameters, which probably factors in to the high CGennFF scores.
© CHARMM forums