Support MDAnalysis.Universe as the frame parameter

[GardevoirX]

Closes #358

This PR allows passing a Universe object to the frame parameter of chemiscope.show. It can be checked by running

import chemiscope
import MDAnalysis as mda

u = mda.Universe("./python/examples/data/trajectory.xyz")
chemiscope.show(frames=u, mode="structure")

As mda is usually used in analyzing biosystem-related trajectories, I also tried it on a pdb file containing two peptides, but I found the current representation schemes cannot display the protein well (see below), maybe we can later consider supporting the ribbon diagram representation?

[Luthaf]

Thanks a lot, this is very nice!

From the screenshot, it looks like the structure is using the wrong unit. Chemiscope/3Dmol.js assumes Angstroms, but I think MDA is using nm. I think there should be at least some automatically guessed bonds in this structure.

As a related point, does MDA add back the implicit H atoms that are often missing in bioinformatics file formats?

maybe we can later consider supporting the ribbon diagram representation?

This is #127, PR welcome! The biggest issue there will be that we will need to add another field to chemiscope's JSON containing information about alpha carbons / some level of pre-computed secondary structure information.

[GardevoirX]

From the screenshot, it looks like the structure is using the wrong unit. Chemiscope/3Dmol.js assumes Angstroms, but I think MDA is using nm. I think there should be at least some automatically guessed bonds in this structure.

In this screenshot, I turned off the bond display, otherwise it would be totally a mess 👇 But I agree that there might be a unit issue. I will look at it later.

As a related point, does MDA add back the implicit H atoms that are often missing in bioinformatics file formats?

No, it doesn't.

[GardevoirX]

Regarding the unit issue, I think mda read the coordinate in the structure file as-is. The real difference comes from the units that are used by different formats (like Å for xyz and nm for pdb, generally speaking). And it might be hard to tell which unit is being using from chemiscope side.

[Luthaf]

https://userguide.mdanalysis.org/stable/units.html Looks like everything should be in A. I'm not sure what's happening in the initial screenshot though. Which file did you use for it?

[GardevoirX]

Ahh yes, thanks! I used this file. And in the initial screenshot, I toggled off the bond. Perhaps you can take the latter one for your reference.

[PicoCentauri]

What is the status here? Would be still nice to have this feature if possible!

[PicoCentauri]

Once this is merged we can add chemiscope here:

https://www.mdanalysis.org/pages/mdakits/#visualization-tools

[PicoCentauri]

Nice. Instead of passing a universe I would pass an Atomgroup, which can be accessed by universe.atoms. In your code you can get the trajectory by atoms.universe.trajectory.

If we allow passing an atomgroup users can easily subselect atoms.

[PicoCentauri]

Nice. The tests are not happy though.

[GardevoirX]

Nice. The tests are not happy though.

Now good👍

[PicoCentauri]

Looks good to me. But maybe also we should add some info somewhere in the docs. Probably @Luthaf can show the correct spot.

[GardevoirX]

Looks good to me. But maybe also we should add some info somewhere in the docs. Probably @Luthaf can show the correct spot.

Maybe we can add something like this example? but if so, I'd like to merge this PR first and add an example in the following PR about introducing cartoon

[Luthaf]

Adding an example later for docs sounds good, although we could update the docstring for create_input, write_input, show (and anything else taking frames) to mention both stk and mda (they currently say "For now, only ase.Atoms objects are supported").

[GardevoirX]

Adding an example later for docs sounds good, although we could update the docstring for create_input, write_input, show (and anything else taking frames) to mention both stk and mda (they currently say "For now, only ase.Atoms objects are supported").

I think that's good for finishing this PR. Adding an example can be postponed, but yes it's nice to show others what we can do through chemiscope