Efficient QM/MM Modeling of Enzymatic Reactions Combining PathCV with OPES.

Hybrid quantum mechanics/molecular mechanics (QM/MM) simulations combined with enhanced sampling are powerful tools for studying enzymatic reaction mechanisms, but their application remains limited by the short time scales accessible at the QM/MM level and by the difficulty of defining efficient collective variables. Here, we present a practical protocol that combines a path collective variable (PathCV) with the exploratory variant of on-the-fly probability-enhanced sampling (OPES) to accelerate reactive transitions and reconstruct free energy profiles in enzymatic QM/MM simulations. A PathCV is constructed from a preliminary reactive trajectory and then used as a reaction coordinate in a subsequent biased simulation, avoiding costly on-the-fly path refinement. We also introduce a block-selection strategy that isolates statistically reliable trajectory segments to enable robust free energy reconstruction from time-dependent biased trajectories. A systematic analysis of OPES parameters further identifies practical deposition and adaptive sigma stride settings. We apply this protocol on three mechanistically distinct enzymes: polysaccharide lyase Alg7A, human -GlcNAcase, and SARS-CoV-2 main protease. In all cases, PathCV-guided OPES promotes multiple reactive transitions within QM/MM-accessible simulation times and outperforms alternative biasing strategies (OPES, metadynamics, and well-tempered metadynamics) in terms of sampling efficiency. Overall, our work provides a practical guide for computationally demending enhanced sampling QM/MM studies of enzymatic reactions.