Иерархическая оценка структурного сходства фармакологически активных соединений

A new method of assessment of the structural similarity of pharmacologically active substances has been developed. The method is based on a hierarchical description of the chemical structure by means of QL-language descriptors with varying degrees of complexity. It is shown that the proposed QL-similarity coefficient can recognize very structurally similar compounds. In silico search for Maillard reaction inhibitors among new adamantane derivatives was carried out by the method of similarity to standards using IT Microcosm software complex. Based on the results of in vitro experiments, two compounds with activity above that of the reference substance (aminoguanidine) and, in addition, seven compounds comparable in activity with aminoguanidine were found.