I was able to use Voidoo which I already had installed from USF to better define the cavity I was interested in. I then was able to use the free web server Dockblaster (http://blaster.docking.org/start.shtml) to screen various sections of the ZINC database and give me some hits. The trick was to position a phosphate in the middle of the site defined by voidoo and then give the coordinates of this "fake" phosphate to Dockblaster in order to define the cavity for the subsequent docking.

I won't take the results of the in silico docking too seriously as I have had limited success with similar methods in the past (and the molecules in the ZINC library are not quite as available as the library advertises), however it does at least give me a start in thinking about the sort of molecules that might fit my cavity.

Thanks once again for the help,

Simon