eSEN-omol MLIP Delivers Quantum-Accurate Full-Enzyme Simulations at 1000x QM/MM Speed

The study applied eSEN-omol to three mechanistically distinct enzymes in explicit solvent, eliminating the QM/MM interface artifacts and region-selection sensitivity that have historically limited hybrid methods. Simulations captured the full Claisen rearrangement trajectory in chorismate mutase, identified a previously unresolved acyl-enzyme intermediate in PETase depolymerization, and distinguished inline versus dissociative phosphoryl-transfer pathways in NDP kinase. Each case matched experimental kinetic trends while running three orders of magnitude faster than DFT-based QM/MM.

Traditional QM/MM workflows require expert partitioning and suffer from boundary errors that scale poorly with system size. eSEN-omol removes those constraints, enabling microsecond-scale sampling of entire solvated enzymes at near-DFT accuracy. This shift matters for drug discovery because many therapeutic targets involve induced-fit or solvent-mediated steps invisible to shorter or fragmented simulations.

The main limitation is the absence of direct experimental validation of the newly resolved intermediates; the paper relies on agreement with existing kinetic data. Larger-scale benchmarking against stopped-flow or cryo-trapped structures would strengthen causal claims. Future work will likely test transferability across enzyme families and integrate the potentials with alchemical free-energy methods for ligand design.