[gmx-users] "VdW radii" of carbons in an unified-atom system

[Kyeong Hyunho]

Hello, dear gromacs users,

I am trying to get a hint of appropriate values of "VdW raii" of carbons.
For this I calculated r_m from gromos53a6. It uses LJ combination rule
#1, so to my understanding the ratio of the last two columns in
ffnonbonded.itp equals sigma^6. From r_m^6 = 2*sigma^6 I could
calculate r_m for each atomtype.
Now I'm confused because the calculated values for r_m are smaller for
HC than C. Moreover they show a monotone pattern; CH1 > CH2 > CH3 >
CH4.

;name, r_m, epsilon
C      0.2010   8.8770
CH0    0.3726   0.2239
CH1    0.2817   3.0365
CH2    0.2284  13.1374
CH3    0.2103  27.7488
CH4    0.2082  40.4358
CH2r   0.2220  15.3369
CR1    0.2100  16.0851
HC     0.1332   3.7882

In combination rule #1, the optimal distance of any two atoms is to be
calculated as the geometric mean of r_m, isn't it? Then these values
indeed roughly translate to sizes. But for instance in phenylalanines,
does HC really pack so much closer with other atoms than C does?

Many thanks,
Hyun-Ho