The overarching goal of this proposal is to computationally model biomolecular binding, iteratively informed by experiments, to fully understand molecular recognition and binding mechanisms. We will apply hidden free energy barriers to modify inhibitors for preferred binding kinetics and use the free energy landscape to understand the role of waters and how and why residues far from ligand binding site can contribute to mutation effects and ligand selectivity. Non-covalent molecular recognition plays a crucial role in biology, chemistry and medicine. Kinetic binding rate constants, together with equilibrium constants, affect the speed, efficacy, and safety of non-covalent drugs and inform their design. In some cases, binding kinetics are the major determinant of a drug’s in vivo efficacy. However, kinetic behavior of ligand binding/unbinding is mainly governed by transient unseen intermediates, which are very difficult to observe experimentally. Computer simulations offer an alternative solution, both for describing and understanding experimentally unseen phenomena and to inform drug design. Real molecular systems are complicated and flexible and call for new modeling tools and theories to compute ligand binding/unbinding free energy profiles. Used in combination with experiments, our new modeling approach integrates data and interprets experiments as a precursor to designing molecules with preferred binding kinetics/affinities. Guided by excellent results obtained during the previous funding period, three Specific Aims are proposed: 1) Develop and apply methods to understand mechanisms and processes of molecular recognition that provide a comprehensive picture and applications for drug design; 2): Understand the binding/unbinding free energy profile from multiple pathways and investigate the effects of waters and sidechain mutations during recognition; 3) Adapt and apply the new methods to ligand binding specificity and kinetics to understand off-site kinase targets. The approach is innovative in its focus on control of kinetic behavior, advanced methods to realistically model free energy profiles and, based on this realism, expand on the classical view of molecular recognition. The proposed research is significant because it comprehensively models free energy profiles, kinetic behavior, detailed water effects, and mutations that may confer drug resistance. Significant outcomes: New computational tools to realistically design ligands with preferred binding kinetics, understand solvent and mutation effects, explain drug selectivity.