Multi-agent scientific research system for predicting dependence "structure-propertie" of artificial immune systems

The article deals with the issues of creating multi-agent Smart-system for conducting scientific research for computer molecular design of new drugs with specified properties and prediction of the „structure-property11 relationship (QSAR) based on modified algorithms of artificial immune systems and other bio-inspcctcd approaches of artificial intelligence. The main advantages and disadvantages of using various intelligent algorithms when building a Smart system arc given. One of the problems with computer-aided molecular design of drugs is the „paradox of similarity11, when the compounds differ structurally quite insignificantly, but have completely different properties. For example, an optically active substance and its mirror isomer can vary significantly in biological activity. Therefore, it is particularly relevant to develop new non-traditional approaches of artificial intelligence and algorithms that provide the ability to recognize chemical compounds with almost the same structure, but with completely different properties. The structure of a multi-agent Smart-system for conducting scientific research has been developed based on modified algorithms of artificial immune systems and the functioning of agents has been described. The work was carried out according to the grant of the Committee Science of the Ministry of Education and Science of the Republic of Kazakhstan (2018-2020), on the topic „Development and analysis of databases for the information system of predicting dependence „structure-property11 of drug compounds based on artificial intelligence algorithms