RMSD and Rotational Constants Modules

[crjacinto]

Closes Issues

Closes #220 #219

Description

• Class methods to calculate the RMSD between two structures and the rotational spectrum (rotor type, moments of inertia and rotational constants) of a molecule.

Release Notes

Added

• Added rmsd and rmsd_kabsch to calculate RMSD without and with rotational alignment to Structure class (RMSD and Rotational Constants Modules  #230).
• Added calc_rotational_constants to calculate molecular rotational constants to Structure class (RMSD and Rotational Constants Modules  #230).
• Added calc_rotor_type to classify a molecule's rotor type to Structure class (RMSD and Rotational Constants Modules  #230).

[timmyte]

Solid first PR.
But the tools are bit to decoupled from OPI. You should use and add to the data structures that we already have in OPI.

[timmyte]

As discussed, I would design the RMSD function as follows:

class Structure:
    def rmsd(self, other: Structure, /, only_atoms: Sequence[int] = (), *, ignore_hs: bool = False) -> float:
        return _rmsd(self.coordinates, other.coordinates, only_atoms=only_atoms, ignore_hs=ignore_hs)

    def _rmsd(coords1: NDArray[float], coords2: NDArray[float], /, only_atoms: Sequence[int] = (), *, ignore_hs: bool = False)
        # We ignore `ignore_hs` if `only_atoms` is populated. This should also be documented
        if coords1.shape != coords2.shape:
            raise ValueError

    def rmsd_kabsch(self, other: Structure, /, only_atoms: Sequence[int] = (), *, ignore_hs: bool = False) -> float:
        # 1) Translate to center of geometry (centroid)
        # 2) find optimal rotation

        # `only_atoms` and `ignore_hs` don't need to be passed here anymore as `coords1` and `coords2_rotated` should only contain the coordinates of the relevant atoms.
        return _rmsd(coords1, coords2_rotated)

[timmyte]

Very nice improvement of the code. We are getting there.
I really appreciate how well you documented the code.

[timmyte]

Thanks for tackling my remarks. There are still some unresolved issues though.

Please also see the few additional points I raised.

[timmyte]

Looks good. Please consider my last remark before merging.
I also would like to get @haneug opinion on the API with all the different weight parameters

[timmyte]

Very good, especially the exhaustive amount of tests. BRAVO!

[timmyte]

Thanks for quickly tackling my remarks. But please make sure you addressed all of them.

[haneug]

Please address all of the comments by Tim and resolve them when they are resolved. I added some more. Also in PR message on its own you should only add to the section Added and do not describe stuff that changed within the PR.
Edit: If they are just regular comments and you cannot resolve them thumps up also works for me to see that you addressed them.

[haneug]

Really nice PR! I just made some small style changes and will merge this now.

Handed this to HN for review