Visualizing the energy landscape for a molecular dynamics trajectory

Abstract

We introduce an open-source program that converts molecular dynamics trajectories into disconnectivity graphs, providing a concise and interpretable visualization of the energy landscape that has been traversed. Our approach applies Savitzky–Golay smoothing to per-frame thermodynamic traces (potential energy in NVE/NVT or enthalpy in NPT ensembles) to identify local extrema as proxies for minima and transition states, and generates the necessary files for disconnectivity graph construction. This workflow requires no additional geometry optimization. The method is ensemble-agnostic and compatible with both all-atom and coarse-grained simulations. For some representative biochemical systems, it processes 104–105 frames in seconds on a standard laptop and produces an approximate representation of the underlying landscape topology. The resulting graphs capture the structures that are visited and pathways between them for a selected energy and time resolution, offering an interpretable structural summary of conformational hierarchies with minimal postprocessing. Because extrema are detected directly from the trajectory, the graphs reflect the organization of the explored region of the landscape on the molecular dynamics timescale. Our approach basically substitutes local minima and maxima from the smoothed time series as proxies for the true stationary points of the underlying landscape.