[gmx-developers] SASA / VDW radii

Berk Hess gmx3 at hotmail.com
Tue Aug 17 09:56:47 CEST 2004

>is anyone aware of gromacs applications that rely on certain values of
>the Van der Waals radii (in vdwradii.dat)?
>The radii are read by the following programs:
>I have a new set of values from the msms program (scripps) with which we
>can reproduce the areas calculated by msms with a different algorithm.
>The differences in radii are quite small, but it would, e.g. give fewer
>solvent molecules when running genbox, and obviously it would also give
>different SASA.

>Values are:
>Atom	OLD (gromacs)	MSMS
>C     0.15		0.174
>F     0.12		0.14
>H     0.04		0.12
>N     0.110		0.154
>O     0.105		0.14
>S     0.16		0.18
>Do you think it is a problem to replace the old values by the ones from

g_sas does not use these radii. The radii are hard coded in gmx_sas.c, see 
I don't know where those come from, maybe from the reference that g_sas 

The MSMS radii are a lot larger than the old ones.
genbox might give much more empty space between solute and solvent,
requiring more equilibration and resulting in a smaller box.
Also gmxcheck -c might complain a lot even when a correct conformation is 
Have you check these things?


real calc_radius(char *atom)
  real r;

  switch (atom[0]) {
  case 'C':
    r = 0.16;
  case 'O':
    r = 0.13;
  case 'N':
    r = 0.14;
  case 'S':
    r = 0.2;
  case 'H':
    r = 0.1;
    r = 1e-3;
  return r;

MSN Search, for accurate results! http://search.msn.nl

More information about the gromacs.org_gmx-developers mailing list