Machine Learning-Based Two-Dimensional Ultraviolet Spectroscopy for Monitoring Protein Structures and Dynamics

Two-dimensional ultraviolet (2DUV) spectroscopy is an emerging spectroscopic technique that offers high resolution and detailed insights into protein structures. However, traditional theoretical calculations of 2DUV spectra for proteins are computationally expensive due to their complex and flexible structures. In this study, we developed a machine learning (ML)-based approach for the rapid and accurate prediction of protein 2DUV spectra. The results demonstrate that, compared to traditional one-dimensional ultraviolet (1DUV) spectroscopy, 2DUV spectroscopy provides higher resolution structural characterization and effectively monitors dynamic processes such as mutations, aggregation, and protein folding. This approach not only offers a cost-effective ML-based solution for predicting 2DUV spectra but also serves as a powerful tool for studying protein structures and dynamics, with potential applications in understanding mechanisms and regulating functions.