In Silico Receptorome Screening of Antipsychotic Drugs

The recent availability of a complete interaction matrix between 13 antipsychotic drugs and 34 protein targets (Roth et al. Nat. Rev. Drug Discov. 2004, 3, 353-359) allows to assess the performance of computational methods on their ability to anticipate the entire affinity profile of drugs across multiple targets. The analyses reveal that our current implementations, based on the similarity of drugs against a reference set of small molecules for which pharmacological data is available in the public domain, are able to predict 65% of the 442 affinities within 1-log unit error, with a level of precision above 92%. In spite of the relatively small scale of this validation study, the results are indicative that in silico receptorome screening of drugs offers an efficient and cost-effective complement to in vitro screening.