Dynamism in multiple selection choices

[kimjoyc]

Right now, I am using SolvationAnalysis to make the shell selection given a frame input:

Then for that given frame I will output this based on solute_idx = 603.

Since frame 0 has ~15 solute indices (i.e. 15 solvation shell selections) and I am loading it directly into add_attribute() as boolean selections.

Here are the selections below:

Example structure ea_fec.

Enable multiple selections for specified solvation shell. Would this be a helpful feature?

[BradyAJohnston]

I haven't done any solvation analysis myself, but it seems like this would be a good feature for people who wish to do so.

I am unfamiliar with solvation analysis myself, but it seems that it would be on a frame-by-frame basis that the selection would be applied. Rather than having to re-import the selection for each frame that you wish, you could apply it to the frames themselves as an attribute:

MolecularNodes/MolecularNodes/md.py

Lines 255 to 271 in 4c19d04

	for ts in traj:
	frame = create_object(
	name = name + "_frame_" + str(ts.frame),
	collection = coll_frames,
	locations = univ.atoms.positions * world_scale
	)
	# adds occupancy data to each frame if it exists
	# This is mostly for people who want to store frame-specific information in the
	# b_factor but currently neither biotite nor MDAnalysis give access to frame-specific
	# b_factor information. MDAnalysis gives frame-specific access to the `occupancy`
	# so currently this is the only method to get frame-specific data into MN
	# for more details: https://github.com/BradyAJohnston/MolecularNodes/issues/128
	if add_occupancy:
	try:
	add_attribute(frame, 'occupancy', ts.data['occupancy'])
	except:
	add_occupancy = False

This is the loop in md.py where I create the frames of the trajectory that act as coordinates. Each frame in the trajectory will have the same number of vertices / atoms as the actual object itself. To animate through the trajectory, I use the frames as a lookup table based on their index and get the position to then interpolate through.

The same principle can be applied to the solvation shell selection, or really any frame-by-frame selection. If you have a new add_attribute() inside of this loop and store the attribute on the frames themselves, then we can just retrieve the selection based on the currently active frame of the trajectory.

for ts in traj:
        frame = create_object(
            name = name + "_frame_" + str(ts.frame),
            collection = coll_frames, 
            locations = univ.atoms.positions * world_scale
        )
        
        if solvation_shell:
            try:
                add_attribute(frame, 'solvation_selection_name', solvation_selections[ts.frame])
            except:
                solvation_shell = False
        
        # adds occupancy data to each frame if it exists
        # This is mostly for people who want to store frame-specific information in the 
        # b_factor but currently neither biotite nor MDAnalysis give access to frame-specific
        # b_factor information. MDAnalysis gives frame-specific access to the `occupancy` 
        # so currently this is the only method to get frame-specific data into MN
        # for more details: https://github.com/BradyAJohnston/MolecularNodes/issues/128
        if add_occupancy:
            try:
                add_attribute(frame, 'occupancy', ts.data['occupancy'])
            except:
                add_occupancy = False

I'm not sure how the structure of the solvation shell is, you have have to use something like np.isin() and the indices to turn it into a boolean selection, but once that is applied on a frame-by-frame basis, then the one selection for the entire trajectory can be imported at the same time.