Is it valid to run gmx_MMPBSA on trajectories with dissociation or migration events?

[Umut Ç.]

Hello,

I am studying the binding free energy of 5 peptide ligands against the same target protein. Each system was simulated in 5 independent replicas of 500 ns each (25 trajectories total). Upon visual inspection and analysis, I noticed that in some replicas the peptide undergoes dissociation, migration to a secondary binding site, or significant reorganization during the simulation.

My main question is whether it is valid to run gmx_MMPBSA on these trajectories as they are.
If running on the full trajectories is not appropriate, I would appreciate any guidance on the following:

1. Is it valid to use startframe/endframe to select only the stable-binding portion of a trajectory, or does this introduce bias?

2. Should replicas where dissociation/migration occurs very early (e.g., within 50–100 ns) be excluded entirely?

3. When comparing across ligands, is a consistent time window preferred, or is it acceptable to use different stable-window lengths per replica?

Any guidance would be greatly appreciated.

Thank you.

[marioe911116]

In most studies, the stable portion ("equilibrated") of the trajectory should be used to calculate the energy. However, if you are more interested in this case in studying the binding-dissociation process, gmx_MMPBSA can also help with this. In short, it depends ultimately on the biological question you want to answer