Difference between x2top and LigParGen ff generation (dodecane)

User discussions
, ,
1

GROMACS version: 2018
GROMACS modification: No
Hi all,

I am trying to run MD simulations of dodecane and currently I’m comparing topologies generated by LigParGen and x2top (given the .gro file generated from the dodecane SMILES provided to LigParGen). I am having a few difficulties.

Firstly, there is a small net charge on the topology generated by LigParGen which for some reason has the charge to 4 decimal places whilst the charge from x2top is less precise and sums to zero as expected. I assume I can use the x2top charges in the LigParGen topology? I know the charge needs to be corrected to net zero, is this an acceptable method by anyone’s experience?

My second issue I’m more confused about. I am getting quite difference bonded, dihedral and angle parameters from both methods. For example, from LigParGen in the bonds section:

[ bonds ]
2 1 1 0.1529 224262.400

and the x2top:

[ bonds ]
; ai aj funct c0 c1 c2 c3
1 2 1 1.520000e-01 4.000000e+05 1.520000e-01 4.000000e+05

so the b is similar but force constant is almost two times higher? I think these should be the same units as well?

and the angles (same order):

[ angles ]
; ai aj ak funct c0 c1 c2 c3
1 2 3 1 112.700 488.273

[ angles ]
1 2 3 1 1.110000e+02 4.000000e+02 1.110000e+02 4.000000e+02

Actually, in the angles section, c1 is always 400 for x2top but for LigParGen it varies between 276.144 and 488.273.

Has anyone else experienced similar/ have an explanation for this behaviour. It was my understanding that I’m using OPLS-AA in both cases so why such a large discrepancy?

Thanks.

2

If you read the gmx help x2top you will see that it is a very primitive tool. The bonds/angles/dihedrals force constants are given as input parameters. If I understand it correctly the bond lengths and angles are set based on the input coordinates.