Project Summary The scientific goals of the funded parent project of this proposal (R35 MIRA for ESI) include the development of global probabilistic and computational models of biomolecules that characterize and quantify the landscape of protein variability and their interactions. The ultimate goal is to elucidate the landscape of functional mutations, which is hidden within the much larger non-functional space. We are using this functional landscape to engineer hybrid transcriptional regulators as well as to predict specificity networks in two-component systems. Another important goal is to devise models to characterize the sequence dependance on protein-protein and protein-RNA interactions. These models will allow us to encode and predict recognition from inferred landscapes and to integrate our results with experimental technologies. Devising the spectrum of functional biomolecular variability sculpted by evolutionary processes will be used to estimate the effects of mutations in disease, antibiotic resistance, biomolecular sensor design and the impact of sequence composition on interaction networks. As the research program of the parent grant benefits from generative modeling and atomistic molecular simulations, we have turned our attention to build a framework that will produce more accurate estimation of the sequence statistics of biomolecules and perform molecular dynamics in an efficient way. For this reason, we request additional funds to build a new Graphical Processing Unit (GPU) system based on NVIDIA A100 technology and large memory CPU nodes. This technology will accelerate the goals of the parent research program in problems related to machine learning, biomolecular simulation and linear algebra calculations.