Artificial Intelligence-Driven Multiscale Design of Protein-Based Materials

Abstract

Protein filaments, including silk, collagen, and amyloids, are natural building blocks whose hierarchical self-assembly produces adaptive mechanics and diverse functions. Conventional protein engineering has had limited success in resolving sequence-structure-function relationships across length scales. Recent advances in artificial intelligence (AI) now provide a framework for physics-informed multiscale design, from molecular motifs to functional macroscopic architectures. Machine-learning models can encode biophysical principles and guide the generation of protein-based materials. When combined with high-throughput experimentation, generative AI can integrate sequence design, structure prediction, and functional validation into closed-loop workflows. These workflows can accelerate the development of programmable biomaterials. This review summarizes AI-driven advances in protein-based materials and emerging links with autonomous robotics.