AI-driven small molecules drug discovery

Artificial intelligence (AI) helps solve problems in various fields, including drug discovery, through its capability to process and analyze large amounts of data in a short period of time. AI-aided drug screening is less time-consuming and less expensive compared to experimental trial-and-error-based screening. Using an AI-based framework that learns the relationship between the structures of small molecules and their biological properties offers new perspectives on molecular design. This presentation introduces the use of multiple machine learning models to predict biological activities based on molecular representations, such as molecular fingerprints and line notations of molecular structures. This method is highly effective for molecular design across many targets. The accuracy of the models was assessed using error metrics and R-squared values relative to experimental IC50 values. The models were then applied to virtual screening to identify top candidates, and their binding affinities to viral protein targets were validated using molecular modeling techniques.