From proteins to nanoparticles: domain-agnostic predictions of nanoscale interactions.

The accurate and rapid prediction of generic nanoscale interactions is a challenging problem with broad applications. Much of biology functions at the nanoscale, and our ability to manipulate materials and engage biological machinery in a purposeful manner requires knowledge of nano-bio interfaces. While several protein-protein interaction models are available, they leverage protein-specific information, limiting their abstraction to other structures. Here, we present NeCLAS, a general, and rapid machine learning pipeline that predicts the location of nanoscale interactions, providing human-intelligible predictions. Two key aspects distinguish NeCLAS: coarsegrained representations, and the use of environmental features to encode the chemical neighborhood. We showcase NeCLAS with challenges for protein-protein, protein-nanoparticle and nanoparticle-nanoparticle systems, demonstrating that NeCLAS replicates computationally- and experimentally-observed interactions. NeCLAS outperforms current nanoscale prediction models and it shows cross-domain validity. We anticipate that our framework will contribute to both basic research and rapid prototyping and design of diverse nanostructures in nanobiotechnology.