Bug in environment selection for dynamic environments

[pranjalsur]

Hi. I was selecting a region geometrically as my environment; molecules moving inside the region should act as my environment (excluding SOL and ions). Turns out , the environment selection is not dynamic in the current code. I have verified this (using the -dumptime flag) by investigating the environment.pdb for several frames. I have fixed this locally by modfiying the line ( in TUPA.py) : env_raw_selection = u.select_atoms(config.sele_elecfield), to "env_raw_selection = u.select_atoms(config.sele_elecfield, updating=True )".

Let me know, if there are any other concerns regarding the fix. If not, the main code should be updated.

[mdpoleto]

Hello! Yes, the environment selection (sele_environment) was not design to be dynamic. Instead, the idea is to have the same group of atoms exerting the field for every frame. In doing so, we can compute correct statistics for residue contributions throughout the trajectory. If residues are not fixed, we can't compute these quantities that are quite useful.

The dynamic selection for atoms in and out of the cutoff was relegated to the Solvent selection (solvent_selection), since solvent molecules tend to be much more dynamic and may leave the region of interest. You may hack this selection to include solute molecules if you want, but they won't be printed in the residue contributions output file.

Thus, I don't really see this as a bug.

[pranjalsur]

Hi. I was not aware of the fixed nature of environment selection, and the results showed discrepency, that's why I used the phrase bug.

I think that making the selection dynamic just makes the tool more generalised. If your intention is to calculate the contribution always from the same group of atoms, setting update = True in environment selection will not change the scenario if the selection keyword is correctly implemented. Residue wise contribution will also be printed.

As I wrote, I aimed to select a region geometrically as my environment; molecules moving inside the region, in presence of external electric field, should act as my environment, be it protein, waters or ions. I want to calculate field at a given point. I can't hack just the solvent selection for this purpose, as then environment becomes empty, code stops working. I include the water or ion in solvent selection, and include protein in environment selection. That's where dynamic environment selection helped.

Although problem can arise: If the dynamic environment is empty at a certain point of time, code can stop working.

Suggestion: If at any point environment selection is empty, but solven selection is not, the code must warn the user, but should not stop the calculation. It is up to the user to address the warnings, if required, based on the warning log file.