[gmx-users] question about g_sas

[Justin Lemkul]

On 5/5/15 11:50 PM, Christopher Neale wrote:
> This is actually kind of unfortunate. I'm not sure where these gromacs VDW radii come from, but they're nowhere near what I presume to be the standards (either Bondi et al. or something similar). Note that even the recent updates to the Bondi values are very minor in comparison to the differences to the gromacs vdwradii.dat values.
>
> I don't know if vdwradii.dat has been updated recently, but my version of vdwradii.dat contains the text:
>
> ; Very approximate VanderWaals radii
> ; only used for drawing atoms as balls or for calculating atomic overlap.
>
> which is quite wrong if they are also used for computing SASA. Looks to me like something that was originally devised as a quick hack for genbox (where precise radii don't matter as much) is being used in code where radii actually matter quite a bit.
>

The most recent version of vdwradii.dat uses the radii of Bondi.  The change was 
actually made back in 2013, but I think it has only been incorporated in the 
release-5-0 branch, so 4.6.x and prior will use the old values.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

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