Computational Approaches in Drug Design

 Drug design is a complex multistage process of finding novel compounds for treatment of a certain  disease based on the knowledge of a biological target. Due to high failure rates, significant research  costs, and technological complexity, drug discovery is an extremely expensive and time-consuming process. Streamlining the early stages of drug discovery can facilitate the acceleration  and cost reduction of drug discovery campaigns. Computer-aided drug design (CADD) approaches have proven efficient in guiding the identification and optimization stages of drug design projects. Due to advances in computing, expansion of virtual libraries of small molecule compounds, and  the growing availability of 3D structural data of biological targets, computational approaches are  increasingly utilized in drug design efforts by both academia and the pharmaceutical industry. One  of the promising computational strategies is the calculation of the pharmacological properties of  interest using computational chemistry methods such as molecular dynamics (MD) and quantum  mechanics (QM). Another efficient approach is leveraging machine learning models to predict the  quantities of interest. The following thesis describes six research studies on the development,  validation, and application of CADD approaches. Chapters 1-5 are focused on MD-based protein?ligand binding free energy (BFE) calculations including the computational protocol validation  (Chapter 1), development of the workflow for ML-guided BFE calculations for molecular  optimization (Chapter 2), the application of the developed protocols in a drug design challenge  (Chapter 3 and 4), and the approach for BFE calculation for a challenging target presented by a  large protein with large flexible binding pocket (Chapter 5). Chapter 6 describes a computational  approach to predict blood-brain barrier permeability using solvation energy calculations with QM  and molecular mechanics methods