BPS2026 – MDAnalysis as a universal connector for molecular dynamics simulation data

MDAnalysis (https://www.mdanalysis.org) is a widely used open-source Python library that provides a unified interface to particle-based simulation data across biomolecular science, soft matter, and materials research. By supporting more than 50 file formats, including remote and cloud-based storage, MDAnalysis decouples analysis workflows from specific simulation engines and offers a stable, interoperable API built on NumPy. Through its converter interface and ecosystem of community-developed MDAKits, it serves as a connector between diverse simulation outputs and advanced analysis tools in the wider Python scientific stack. We present recent developments that extend MDAnalysis beyond file-based workflows to real-time data streaming. Modern all-atom simulations generate petabyte-scale trajectories that are rarely stored in full, forcing coarse sampling and the loss of fast dynamical information. The new streaming interface, implemented for engines such as GROMACS, NAMD, and LAMMPS, enables MDAnalysis to access all data directly as a simulation runs. This approach allows event-driven analysis, selective resolution of different subsystems, and real-time computation of fluctuation-based observables without writing large trajectory files. By bridging diverse data sources—from static files to live simulations—MDAnalysis acts as a universal connector that empowers reproducible, engine-independent workflows for molecular dynamics analysis.