Varlamova K. Prediction of properties of chemical mixtures by QSAR/QSPR approaches based on simplex representation of molecular structure

The dissertation is devoted to the development of descriptors for chemical mixtures and application of this system to the solution of QSAR/QSPR tasks. Descriptors representatiing the mixtures of chemical compounds which were developed on the base of unbounded simplexes. The relationship between the structure of organophosphates and their inhibition effect on acetylcholinesterase was obtained. The possibility of structural representation of racemates at 2D level by means of incorporation of stereochemical labels for unbounded simplexes was demonstrated. Adequate structure-property models of toxicity towards Daphnia magna of sodium salts of alkoxycarbonylsulfuric acids, aliphatic n-alcohols, phenols, and their mixtures were developed. The possibility of prediction of a combining effect of antiviral preparations on inhibition of replication of poliovirus was shown. Promising combinations of antiviral agents including new drugs absent in the modeling set were proposed. The models for prediction of boiling/dew points of single organic compounds and their mixtures were developed. Independent external validation revealed that the obtained models are suitable for prediction of properties of new mixtures formed by the investigated 67 compounds (more than 2000 new mixtures).