ABSTRACT There are ~40 million people world-wide infected by the Human Immunodeficiency Virus Type 1 (HIV-1, commonly referred to as HIV). As currently there is no cure, antiretroviral treatment is the primary treatment option. Yet antiretroviral treatment eventually fails over time due to the development of drug resistance. We will develop new computational tools for forecasting HIV evolutionary trajectories under therapeutic selection pressure leading to drug resistance, using high resolution all atom molecular dynamics simulations together with physics-based machine learning models of sequence co-variation. The computational studies will be complemented by structural, biophysical, and virological studies on two HIV protein multimeric targets: HIV integrase (IN), and capsid (CA). The modeling and experiments will be employed in an iterative manner, with the experimental results being used to validate, parameterize and improve the models; and the molecular dynamics simulations used to guide new experiments and also to develop new tools for high resolution cryo-EM refinement of multiple binding modes of HIV inhibitors and interfacial solvent. The common theme of our proposed work is to provide structural interpretations for the observed fitness and resistance effects of mutations, with the goal of developing holistic structure-function models which can be used to predict viral mutation trajectories under drug selection pressure and give a mechanistic explanation for them. There are three specific aims: (1) determine the physical mechanisms underlying mutational epistasis under varied drug environments, and use MD simulations and virological data to parameterize drug specific landscapes for HIV IN and CA under a novel theoretical framework; (2) use high resolution, large scale alchemical molecular dynamics free energy simulations based on advanced sampling methods to analyze the effects of protein mutations on the stability of protein-protein interfaces that constitute the intasome and the capsid particle assemblies; (3) determine the molecular basis for multiple binding modes of inhibitors of HIV IN, and the role of solvation in the strong binding of these inhibitors. These aims seek to achieve a molecular understanding of the cooperative effects (epistasis) of multi-residue mutation patterns on the binding of inhibitors to their viral protein targets (IN, CA) and their effects on the stability of the multimers (intasome and capsid particle). We anticipate that this work will lead to the development of surveillance tools to forecast the response of viral systems to the selection pressure of antiviral therapeutics.