QM/MM Geometry Optimisation with Qsite

Chose rightly and truncate wisely!

From our three 5 ps QM/MM production runs, we can summarise the geometrical parameters of ligand-protein interactions. However, to get a representative structure for reactant, transition state and product state we need to optimise the structure further. For this, we choose to take a random frame from our last QM/MM production run, and further truncated it to be used as input for a QM package. The truncation step involves, the removal of solvent molecules those are not within 5 Angstrom from any protein atom. Truncation step can be done by using the graphical structure modelling programs like VMD or Pymol or by using any programming tools like Biopdb module of Biopython library in python. Our main goal is to understand the structure of the active site and the nearby residues, we believe it would be better to leave the solvent molecules far apart from the active site, at the same time, it would save some computation time as well.

How to choose a snapshot from the MD simulations for QM/MM geometry optimisation?

For modelling the structural coordinates of the biocatalysis reactions, we often took random snapshots from the MD simulations, usually more than one for better statistical assessment. We advise a more investigative approach rather than choosing a random snapshot. In our case, we monitor some key geometrical parameters pertaining to hydride transfer, like the distance between the hydride acceptor (N1) and donor (N5), the angle between the N5, N1 and hydride (H1) etc. A table summarising these geometrical parameters is available as Table S1 in Sahrawat et al. (2024). The average value of the these parameters have been considered to obtain a frame from QM/MM MD simulations. This ensures that the chosen structure represents the active site better than a randomly chosen structure.

We have used Qsite (the QM/MM module of the Schroedinger package) for modelling the reaction coordinates using the truncated structure as an input. A brief description of our chosen parameters for QM/MM geometry optimisation is available in the methodology section of the supplementary information of Sahrawat et al. (2024). There are already several tutorials available demonstrating the QM/MM geometry optimisation of reaction coordinates:

QM/MM Geometry Optimisation Tutorials Qsite –> https://www.schrodinger.com/training/defining-qm-and-mm-regions-qsite221 Gaussian –> http://schlegelgroup.wayne.edu/Software/oniomtoolTAO/TAOtutorial.html Orca –> https://www.orcasoftware.de/tutorials_orca/multi/basics.html NWChem –> https://nwchemgit.github.io/ONIOM.html

Our QM/MM geometry optimised structures of the active site for each reaction coordinate are available in our GitHub repo in tutorial/qmmm-geom-opt directory.