[Dock-fans] Updated Sphgen C++ Version

Andrew Magis andrewmagis at gmail.com
Fri Sep 13 16:33:18 PDT 2002


I have translated the sphgen program (provided in the DOCK 5.3.0 suite) from
the original Fortran into C++. Due to changes in how the program reads in
the .ms file and organizes the data (in class objects instead of arrays) the
input problems formerly caused by generating large numbers of surface points
in dms do not occur.  In addition, there are no upper limits on spheres
generated, as all memory is now dynamically allocated.  By reorganizing some
of the calculations and eliminating most of the disk accesses I was able to
achieve an approximate 2.5X speed increase over the original program.
Finally, the code is now much more readable for those who wish to examine
it.

Input to the program is identical to the original, though I plan on adding
command line input to provide an altermative to the INSPH file-based input
now required.

The only current difference in output is the ordering of the cluster spheres
- they are no longer ordered by atom, but are output in an order determined
by the breadth-first search algorithm I use to identify the spheres
belonging to a particular cluster.  If this has any adverse effects on
DOCKing calculations (I haven't examined the dock source code in depth),
cluster spheres could easily be sorted by atom number before they are
output.

I would be happy to provide all my source code free to DOCK developers or
users who wish to examine it or use it, with the permission of the original
authors of the code from which my software is heavily drawn.   I have done
testing, but more testing is always welcome.

My eventual goal may be to unite the dms program, sphgen, and a few of the
helper programs into a single piece of software that largely automates the
process of generating the surface points, spheres, and cluster files for
novice users, while retaining the level of control now provided by the
original software.

I would like to send the code to the DOCK developers if they are interested,
and obtain permission to distribute the code to individual users, if the
code turns out to be acceptable.

Thanks,
Andrew Magis
andrewmagis at gmail.com
Computer Science MS Student
University of Florida
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://blur.compbio.ucsf.edu/pipermail/dock-fans/attachments/20020913/af8367e0/attachment.html


More information about the Dock-fans mailing list