i-gRINN: a next-generation web server for protein energy network analysis of heterogeneous biomolecular systems with natural language-based data exploration

Protein energy networks (PENs)-residue interaction networks weighted by force-field-based pairwise non-bonded interaction energies-provide a physically grounded framework for identifying functionally important residues, allosteric communication pathways, and ligand-induced network rewiring from molecular dynamics (MD) simulation data. We previously developed gRINN (get Residue Interaction eNergies and Networks), a standalone tool for PEN analysis of biomolecular simulation trajectories, which has seen wide adoption but also suffered from compatibility issues with modern MD engines and operating systems. Here, we present i-gRINN (interactive platform for gRINN), a completely redesigned web server for calculation of non-bonded residue interaction energies and PEN analysis of GROMACS trajectories or PDB ensembles. i-gRINN introduces two major advances over the original tool: first, pairwise interaction energy calculations are extended beyond standard amino acids to include small molecule ligands and non-standard residues; and second, the result dashboard integrates an LLM-powered chatbot that allows users to interrogate interaction energy matrices and PEN metrics using natural language queries, with a two-stage biological interpretation pipeline grounded in UniProt annotations and PubMed literature. Interactive visualization is provided through dedicated panels for pairwise energies, the full interaction energy matrix heatmap, network analysis, and an integrated 3D structure viewer. i-gRINN is freely available without any login requirement at https://grinn.bio-cloud.site.