| Abstract Scope |
The development of biologically inspired materials typically involves extensive trial-and-error studies. Rational understanding and design using modeling and simulation become increasingly feasible due to more accurate models and affordable computing resources. We will share atomic-level insights insights into biomaterials properties at the 1 to 1000 nm scale using the Interface force field (IFF), including recognition and assembly of metal, oxide, and biomineral nanostructures mediated by biomolecules and polymers. Examples include nucleation and growth of bone, low dimensional materials, catalysts, hydrogels, and therapeutics. We will then discuss new opportunities using reactive simulations (IFF-R) and data science tools to learn and interpret the information contained in large computational and experimental data sets to accelerate property predictions. We outline the process of generating a feature representation, the translation into reinforcement learning with nodes and edges, and Bayesian-based uncertainty quantification of predicted properties. Requirements for data sets and first applications will be described. |