Computer-aided design of polyketides with the required properties

An approach to rational design of new polyketides with the required spectrum of biological activity has been proposed. We have developed the BioGenPharm software, which generates combinatorial libraries of polyketides based on the user-defined input parameters, performs prediction of biological activity spectra for the generated structures and selection of molecuels with the required properties. PASS algorithm has been applied for prediction of polyketide activity spectra (http://www.ibmc.msk.ru/PASS). Validation of PASS prediction ability for polyketides was performed vs. the evaluation set containing 242 natural macrolides from the Dictionary of Natural Products. The mean prediction accuracy was 75.5%. The problem of choice of the cutting points for probability of the presence of activity (Pa), which provides optimal combination of such parameters as sensitivity, specificity, concordance has been considered. Applicability of the described method has been illustrated by generation of a virtual library of the erythromycin analogues and selection of substances with low probability of the hepatotoxic effect.