[chimera-dev] Calculation voronoi volums in a prot.

[Conrad Huang]

First the bad news.  I looked into qhull and it appears that computing 
the volumes for all Voronoi cells is not built-in to the software.  The 
qhull FAQ gave an example for computing the volume of a single cell:

> To compute the area and volume of the Voronoi region for input site 5 (site 0 is the first one), use
> 
>     rbox 10 D2 | qvoronoi QV5 p | qconvex s FS

My guess is that this will probably be pretty inefficient as the number 
of atoms increases.  As the qhull documentation says, algorithms that 
keep intermediate convex hull results are much more efficient.

On the good news side, the code that the "Area/volume from Web" server 
is from Mark Gerstein (see 
http://www.molmovdb.org/geometry/files/libproteingeometry/README.html) 
and is probably the best starting point.  If you have some programming 
expertise available, it should be fairly straightforward to compile the 
calc-volume program and modify the Chimera extension to use that instead 
of the web.  (We will, of course, be happy to answer any questions.)  If 
not, we can talk about alternative solutions.

Conrad

Jean Didier Pie Marechal wrote:
> [...]
> 
> Re. the server access. How the call to the server works is perfect
> for the purpose of our work and does exactly what we want. However,
> since we want to script every step of our work, it would be great to
> know if a command line equivalent to the "Area/volume from Web"
> interfase exists.  As we thought there was no command to do so, we
> were wondering if qhull based calls could do the trick.
> 
> Re. Voronoi calculation Conrad, having qhull providing volmetric
> values would be great. That would afford us calling the server, eases
> the script, and also allows us to understand better why in some
> cases, the calculations give strange values (not only the -1 we
> understood the origin, but other problems).
> 
> If that is too complicated, as Elaine says, we will try to call the
> server directly and recover the volumes.
> 
> Thanks a lot for the help,
> 
> Best, JD