Reaction-Coordinate-Guided Energy Landscape Visualization

Proteins explore multidimensional energy surfaces composed of multiple metastable states and kinetic barriers that govern folding and conformational reorganization processes. Characterizing these energy landscapes from molecular dynamics simulations remains a major challenge due to the high dimensionality of the generated data. The Energy Landscape Visualization Method (ELViM) was developed to project high-dimensional conformational ensembles into lower-dimensional spaces using exclusively structural information, enabling the identification of metastable states and multiple transition pathways without requiring predefined reaction coordinates. However, as with other dimensionality reduction approaches, strictly two-dimensional projections may introduce geometric ambiguities and hinder the physical interpretation of the underlying energy landscape. Here, we present a new extension of ELViM, termed the Tomographic Dimension, designed to reconstruct energy landscapes with enhanced physical fidelity. This approach incorporates a physically meaningful reaction coordinate as an additional fixed dimension during the projection process. While this dimension explicitly represents the progression along the conformational process of interest, the remaining coordinates are iteratively optimized to preserve both the original structural relationships and the constraint imposed by the additional coordinate. The resulting three-dimensional representation organizes the conformational ensemble into distinct “tomographic slices,” allowing the distribution of conformational states to be visualized along a physically interpretable axis. The proposed methodology promotes a clearer separation of conformational funnels, improves the identification of metastable states, and reduces ambiguities inherent to conventional two-dimensional projections. Furthermore, it enables a more intuitive reconstruction of the energy landscape topology, bringing computational visualization closer to the classical folding-funnel framework. The method has been implemented as an open-source, platform-independent tool, broadening its applicability to the investigation of complex biomolecular processes, including protein folding, conformational transitions, and molecular association mechanisms. By integrating structural similarity with physically relevant reaction coordinates, this framework provides a powerful and general strategy for the visualization and interpretation of multidimensional biomolecular energy landscapes.