[gmx-developers] SAS or SES algorithm in GROMACS

[Martin Höfling]

Am Dienstag, 23. Januar 2007 20:47 schrieb H.J.Risselada:

Hi Jelger,

> This might have been a bug or simply a limitation in the
> approximation of the surface. The "ball roll" or the
> "sphere overlab" algorithm have diffirent limitations.
> For some proteins (the more glocular onces), however,
> g_sas will probaly do a better job then for others.

Sure, there is probably no perfect algorithm. I observed the same that you 
described here. The only other option for me now was msms which is fast but 
tends to be unstable. This is no problem if you're going to draw a surface of 
one pdb file or so but especially when you're extracting data from 
trajectories, a segfault on frame 599 of 600 is really annoying ;-)

> Ofcourse it will also have an affect on the values of the
> solvent accesible surface, although the scaling might
> still be correct if this error is more or less constant.
> Infact as my collegue checked this, g_sas, nacces or VMD
> gave him quite different values for the solvent accesible
> surface but globaly similar scaling.

VMD uses surf and msms if i am right. I checked surf in the last hours and it 
looks good, although the code seems to be written in a really "old fashioned" 
c-style. Performance: I can't tell but the surface grid looks good. I'll also 
ask Warren De Lano which algorithm he uses in pymol. This would also be 
interesting for gromacs cause the code is BSD licenced so it should be no 
problem to relicense relevant parts under GPL. Anyway I'll report my 
experiences here.

Best wishes
	Martin

-- Everything that is really great and inspiring is created by the individual 
who can labor in freedom. (Albert Einstein)