A Dataset of Interaction Energies from FMO Calculations on Representative Protein Structures

The biological functions of macromolecules are closely associated with their three-dimensional structures and corresponding electronic states. Quantum chemical calculations are an approach that can determine electronic states from first principles without relying on empirical parameters. In this study, we present a quantum mechanics-based dataset derived from representative structures curated in SCOP2. The calculations were performed using the fragment molecular orbital（FMO）method, which is applicable to large biomolecules such as proteins. The dataset comprises total energies for approximately 5,000 structures and over 200 million inter-fragment interaction energies, including electrostatic and dispersion components. This resource is expected to support functional analyses based on electronic structure and to facilitate machine learning studies in biomolecular interaction studies.