In article: <Pine.A41.4.05.9912211105220.99006-100000 at node5.unix.Virginia.EDU>
Pornthep Sompornpisut (ps2t at unix.mail.virginia.edu) wrote:
> I have a few questions concerning about calculation of the solvent
>accessible surface area of proteins. Here they are as follows:
>> 1. Does anyone know if there is any source code available for
>calculating accessible surface area derived from analytical
>function?
> 2. What is the best current approach to express the surface areas as a
>function of interatomic distance?
>
1) Formulas for analytical computations of spheres intersections are
in my paper: J.Comput.Chem.1994,15[5],507-523. The associated executable
programme ASV is available (free) on various platforms, including RS6000,
CRAY J90 and YMP familiy, DEC-alpha (VMS and unix), SGI (various
architectures) and LINUX. Just specify which version(s) you need and
you shall receive it via email. If you need a rapid computation for
large molecules without the analytical accuracy, the Monte-Carlo code
is also included is ASV. The Monte-Carlo subroutines (FORTRAN) are
also free. The full sources of ASV (which computes both surfaces and
volumes) are available for people initiating a collaboration.
2) Surfaces areas are expressable from sphere radii and cartesian coord
of centers, and coord are get from the matrix of interatomic distances
with multidimensional scaling. Recovering N 3-tuples of coordinates
from the N*N distances matrix via multidimensional scaling can be found
in most descriptive statistics textbook. If you have an incomplete
distance matrix, you have to operate iteratively, assuming that the
coord of a 3D point are obtainable from the 3 distances from 3 reference
points together with a flag showing on which side of the plane is the
new point. The analytical formula is also in my paper, because it is
equivalent to look at the two 3D-points at the intersection of three
3D-spheres.
Michel Petitjean, Email: petitjean at itodys.jussieu.fr
ITODYS (CNRS, ESA 7086) ptitjean at ccr.jussieu.fr
