Hey, thanks for the ZINC database it will indeed be usefull but probably later on;-), right now I'm working with four receptors, and I'm trying to identify ligands that have experimentally validated affinity for every one of them (my ultimate goal is to find compounds that bind only to one of those proteins, and are experimentally proven not to bind with other ones, e.g. identify selective ligands)

if you're working with nuclear receptors, i would suggest the nrlist

http://nrlist.drugdesign.fr

http://nrlist.drugdesign.fr

I hop there is no method in docking protocol to screen selective ligand, we can only find interaction and highest binding affinity ligands to specific protein than than specific selectivity

The best way is searching experimental papers on selectivity and your target proteins. That's the only reliable solution which surely is boring and time consuming. Probably a considerable amount of data is lacking. Therefore, after finding a proper number of selective compounds, you can dock them to all of your targets. A rough selectivity index can be calculate as the ratio of binding affinities of the ligand to proteins A and B. Deviation from unity can be a measure of selectivity but since docking is never adequate, you should find an average value for your reference entities. You can then try docking your ligands with unknown selectivity and compare their calculated selectivity indices with your new ligands.

Please read my current publication.

Article In silico structure-based screening of versatile P-glycoprot...

Is your method available online?

Try ligdig  which will find ligands inhibiting respective enzymes.