The ontological models of swarm intelligence algorithms for immune network modeling of drugs

Abstract

The article is devoted to development of information system of forecasting properties of new drugs based on methods of swarm intelligence and artificial immune systems. An In forecasting relationship between the chemical structure of substances and their biological activity (QSAR, Quantitative Structure-Activity Relationship) an important aspect is feature selection. A perspective direction in QSAR is the use of artificial intelligence, which provide high accuracy of forecasting of chemical compounds with predetermined properties. The article considers the methods of bee colonies and particle swarm algorithm for solving the problem of feature selection and further immune network modeling of pharmacological activity of chemical compounds. The existing software of implementation of these algorithms for constructing of optimum set of descriptors are given. Application of multialgorithmic approach in immune network modeling of drugs requires the systematization of using methods and development of an integrated ontological model. Development of ontological models allows to structure input and output data, to consider features of functioning and relationships, saves time and computational resources in the development of component-based software for immune network modeling of new substances with predetermined pharmacological activity. The ontological models of bee colony and particle swarm algorithm for solving the problem of feature selection are created in ontology editor Protege.