Mol. Machines represent a class of proteins that generate force and motion through conformational change and are implicated by genetics in many human diseases.Historical drug discovery against this class of proteins, although generated multiple transformative drugs, are extremely difficult to scale.To solve this problem, we are taking a different approach to systemically study and predict the function, conformations, binding pockets, and relevant chem. for this target class.Here, we describe the hit discovery workflow we have developed against this class of targets based on our in-depth knowledge of conformation-pocket-chem. relationship.From the initial hits, we efficiently navigated ultra-large catalog databases such as Enamine REAL using a combination of ligand- and structure-based virtual screening and machine learning approaches to perform iterative hit optimization and scaffold hopping.This approach resulted in the discovery of several structurally distinct inhibitors with improved potency and selectivity.