Abstract
Noncovalent interactions (NCIs) are fundamental to the structure, stability, and function of proteins. These interactions form complex networks that control how different protein regions relate to each other or to external molecules (e.g., solvent, ligands, cofactors, other proteins), shaping the energy landscape of the system. Molecular dynamics (MD) simulations are widely used to study proteins and other biomolecules, helping to explore their structures and transitions between them at atomic resolution. However, the analysis of MD trajectories is often limited to the estimation of geometric features or metrics relating instantaneous configurations to reference structures, which may overlook relevant details of the interactions that drive conformational changes. In this work, we propose a systematic approach to the analysis of simulation data based on NCIs. We use electron density features from topologically meaningful regions to characterize inter-residue NCIs, computing them with NCIPLOT4 across MD simulations to investigate their presence, conformational relevance, and temporal evolution. This enables a direct, data-driven view of how specific interactions contribute to the stability and rearrangement of structural elements. This allows us to map the interactions shaping protein conformations and how they change along certain processes. We apply this framework to ultralong equilibrium trajectories of protein folding, revealing patterns of interaction changes that correspond to distinct folding pathways. This NCI-based approach provides a powerful complement to the traditional structural analysis toolbox, deepening our understanding of protein folding dynamics.