CHARMM Development Project Forums User Discussion & Questions Parameter Set Discussion Question about Urea parameters (Caflisch and Karplus, 1995)

I took the parameters that Amedeo Caflisch posted to these boards (from his 1995 paper with Karplus) and pasted them into the top_all27_prot_lipid_na file (the exact format is pasted below). I've tried using this file to generate a psf from a pdb file containing urea. However the program (psfgen from the VMD package) doesn't seem able to build urea.

I'm wondering if this is because the parameters will only work with charmm22 or if I haven't added them to the topology file correctly.

Anyone have any suggestions? The information in the topology file is:

============================================================
RESI UREA 0.00 ! urea by A. Caflisch December 19, 1995
GROUP
ATOM O O -0.51
ATOM C C 0.51
GROUP
ATOM N1 NH2 -0.62
ATOM H11 H 0.31
ATOM H12 H 0.31
GROUP
ATOM N2 NH2 -0.62
ATOM H21 H 0.31
ATOM H22 H 0.31
BOND O C C N1 C N2 N1 H11 N1 H12
BOND N2 H21 N2 H22
IMPH C O N1 N2
IMPH N1 C H11 H12
IMPH N2 C H21 H22
PATCHING FIRS NONE LAST NONE

============================================================

Thanks.

I think the dihedrals are missing. I will see to it as I need a molecule of urea as well.

best regard,
zee

PS: I hope it's not to late.

DIHEdrals are not needed in the RTF. As to why psfgen cannot build urea I suggest asking the VMD/NAMD community. Our urea definition is posted below (compatible with top/par_all22_prot):

RTF:
RESI UREA 0.0 ! Based on OPLS Urea. Jorgensen JACS(1993) 115 9271-9275
! Caballe&Nilsson (2005). J. Mol. Struct. THEOCHEM 758:137-146.
GROUP
ATOM N1 NH2 -0.569 ! geometry from Swaminathan et al (1984) Acta
ATOM H11 H 0.416 ! Cryst B40, 300
ATOM H12 H 0.333 ! NOTE: approximation to OPLS, using std ASN types
ATOM C CC 0.142 ! and charges to agree w/ results of Karlstrom et al
ATOM O O -0.502
ATOM N2 NH2 -0.569
ATOM H21 H 0.333
ATOM H22 H 0.416
BOND N1 H11 N1 H12 N1 C
BOND C O C N2 N2 H21 N2 H22
! IMPROPER SETUP TAKEN FROM ASN&GLN
IMPR C N2 N1 O
IMPR N1 C H11 H12
IMPR N2 C H21 H22

ACCE O C
DONOR H11 N1
DONOR H12 N1
DONOR H21 N2
DONOR H22 N2
IC H11 N1 C N2 1.0 119.0 180.0 117.0 1.35
IC H11 C *N1 H12 1.0 119.0 180.0 120.0 1.0
IC O N2 *C N1 1.26 121.0 180.0 117.0 1.35
IC H22 N2 C N1 1.0 119.0 180.0 117.0 1.35
IC H22 C *N2 H21 1.0 119.0 180.0 120.0 1.0
PATCHING FIRS NONE LAST NONE


parameter file:
ANGLES
NH2 CC NH2 130.0 117.0 ! For urea. LNI Sept 1998, analog w/
! NR2 CPH2 NR1, geometry from Acta CrystB40 p300

DIHEDRALS
H NH2 CC NH2 1.4 2 180.0 ! For urea. LNI sept 1998
! analogy with entry for acetamide H NH2 CC CT2
IMPROPER
O NH2 NH2 CC 45.0000 0 0.0000 ! For urea. LNI sept 1998
! analogy with entry for acetamide O HA NH2 CC

Dear Lennart,

I ran two simulations (urea in a water box) with two different force fields for urea (Caflisch's and yours that I found in this forum), and calculated the interaction energy between urea and water using the VMD tool. Here is what I have:

Caflisch's : ~ -30 kcal/mol
yours : ~ -50 kcal/mol

The size of water box is 20 x 20 x 20, and I ran 1 ns for each force field. The difference is huge. I wonder which one is close to an experimental value. Could you give some comments on it ? Thank you.

AFAIK, VMD does not have a force field engine. It is not clear from your post which software you used for simulating the system and calculating energies, what type of simulation you did and what kind of analysis was performed to obtain the values you're citing, leave alone what the meaning of these values is.

Sorry for confusion.
I used NAMD for my simulations. I read in dcd file on VMD and used NAMD energy protocol to calculate the interaction energy between urea and water molecules (I have only one urea in my TIP3P water box).

Thanks.

This would be related to the enthalpic component of the free energy of transfer from gas-phase to aqueous solution, and I don't know if there is an experimental number available for this. The charges are not all that different between the two models though.

It would be interesting to know if you also find differences in the g(r) for water oxygens around the urea oxygen and nitrogens, and in urea translational and rotational diffusion. You may need more than 1 ns, but this is only a question of about an hour/ns of simulation in CHARMM, a little longer probably with NAMD but still manageable.

Thank you for your comments.
I'll look into g(r) and the diffusion.
I wonder whether there is any references for comparison between two force fields.

There are several papers which compare protein force fields, which should be easy to find.

While perhaps not so easy to find*, there ought to be some experiment data for urea available, such as crystal and NMR studies, and maybe transfer free energy or free energy of solvation. In part because of the TIP3P water model, diffusion can be a bit tricky to compare between simulation and experiment. (We have a fairly detailed paper on this in press at JPC B.)

* The common use of urea as a denaturant tends to confound simple search strategies.

About the charges-- they look somewhat different to me, both in magnitude and symmetry assumptions (non-equivalent proton charges). The C=O dipole is also substantially changed.

Yes, the carbon charge is quite different. I hope our old papers on diffusion of TIP3P and comparison with experiment don't come out to badly - if I remember correctly motions in TIP3P are roughly twice as fast as experimentally observed, so it should, to first order, be a relatively straightforward correction. I look forward to reading your paper for more details!