binding pocket?

 Hello all;
 I'm looking for the impossible.
 If I have an inhibitor and a protein that are known to associate, and a set
 of mutations that appear to reduce the in vivo effects of this small molecule,
 is there a way to (reasonably) find a binding pocket for this molecule on the
 protein surface? The normal substrate for the molecule shows little, if any,
 structural similarity to the inhibitor, so the the substrate pocket would
 _not_ seem to be the binding site.
 Does anyone have experience/knowledge of this sort of thing.
         Dr. M. Morgan Conn
         Department of Chemistry (Schultz group), UC Berkeley
         mmconn $#at#$
         510.642.9249; 510.643.6890 (fax)