New Applications of Deep Learning in QSAR

Molecules are the basic units that constitute matter. The structural features of molecules within a compound and the combination mode of atoms determine the properties of the compound. That is, the physical and chemical properties and biological activities of a compound are expressed and explained based on the molecule as the main body. When the molecular structure of a compound changes, its properties and biological activities will also change accordingly. Computer-Aided Drug Design (CADD) is an emerging discipline that applies molecular simulation technology to drug research and development and integrates with traditional pharmacology. It comprehensively uses various theories and computational methods to study the physical and chemical properties and movement behaviors of drug targets or small molecules. After obtaining relatively accurate and detailed physical, chemical and biological information, it guides drug design through the integration, transformation and statistical analysis of the obtained data. Quantitative Structure–Activity Relationship (QSAR) is an important research method in modern drug design, which uses mathematical models to explain the quantitative relationship between the chemical structure parameters of ligands and their biological activity intensity, optimize the structure of ligands and predict the biological activity of new compounds of the same type.