Fast drug-receptor mapping by site-directed distances: a novel method of predicting new pharmacological leads.

The searching and characterization of large chemical databases has recently provoked much interest, particularly with respect to the question of whether any of the compounds in the database could serve as new leads to a compound of pharmacological interest. This paper introduces a fast and novel method of determining whether any of a given series of compounds are able, on geometrical grounds, to interact with an active site of interest. The C program written to implement the method is able to make a qualitative prediction for a given compound in about 1 s per structure (for drug-sized molecules), while still permitting the compound complete conformational freedom. However, the algorithm is sufficiently flexible to permit distance constraints to be placed on the molecules while docking. The test system studied was a family of Baker's triazines docking into the active site of dihydrofolate reductase (DHFR), as defined by a methotrexate/NADPH complex.