repr of TopologyObjects should show order of underlying atom IDs

[orbeckst]

Expected behavior

Representations of topology objects should list atom IDs in the order of the underlying AtomGroup.

Actual behavior

I found it quite confusing when I looked at a Dihedral object and the order of atoms was different from my carefully crafted order to define the dihedral:

>>> dih = u.atoms[[10, 0, 1, 3]].dihedral
>>> list(dih.atoms)
[<Atom 11: HG1 of type 3 of resname MET, resid 1 and segid 4AKE>,
 <Atom 1: N of type 56 of resname MET, resid 1 and segid 4AKE>,
 <Atom 2: HT1 of type 2 of resname MET, resid 1 and segid 4AKE>,
 <Atom 4: HT3 of type 2 of resname MET, resid 1 and segid 4AKE>]
>>> repr(dih)
'<Dihedral between: Atom 3, Atom 1, Atom 0, Atom 10>'

I’d much rather have the actual order output.

Relevant code

Why do we do something special to the order of atom indices in an TopologyObject when printing the repr? Namely in https://github.com/MDAnalysis/mdanalysis/blob/4f36e2a06eb2d04274dfe4510910986c2dd17681/package/MDAnalysis/core/topologyobjects.py#L118

indices = (self.indices if self.indices[0] < self.indices[-1] else self.indices[::-1])

why do we reverse the order?

Code to reproduce the behavior

>>> import MDAnalysis as mda
>>> from MDAnalysis.tests.datafiles import PSF, DCD,  GRO, PDB, TPR, XTC, TRR,  PRMncdf, NCDF
>>> u = mda.Universe(PSF, DCD)
>>> dih = u.atoms[[10, 0, 1, 3]].dihedral
>>> list(dih.atoms)
[<Atom 11: HG1 of type 3 of resname MET, resid 1 and segid 4AKE>,
 <Atom 1: N of type 56 of resname MET, resid 1 and segid 4AKE>,
 <Atom 2: HT1 of type 2 of resname MET, resid 1 and segid 4AKE>,
 <Atom 4: HT3 of type 2 of resname MET, resid 1 and segid 4AKE>]
>>> repr(dih)
'<Dihedral between: Atom 3, Atom 1, Atom 0, Atom 10>'

Current version of MDAnalysis

• Which version are you using? (run python -c "import MDAnalysis as mda; print(mda.__version__)") 2.5.0
• Which version of Python (python -V)? 3.10.0
• Which operating system? macOS

[richardjgowers]

@orbeckst the "forwards" and "backwards" of bonds/angles/torsions are equivalent, so there's a canonicalisation check to make sure that the first atom is less than the last. Internally we often use this check to deduplicate bonds.

WRT this repr, I think I've taken this a little too far and made the repr print out the "canonical" order of the atoms, rather than the true internally stored order of the data. Fixing it seems like a good idea

[ztimol]

Just want to confirm if the canonicalisation check is still required at all? If not we could remove

indices = (self.indices if self.indices[0] < self.indices[-1] else self.indices[::-1])

and simply return the representation in the order of self.indices.

[richardjgowers]

Yeah removing that is harmless and would solve the issue

On Sat, Jul 8, 2023 at 08:53, ztimol @.***> wrote:

Just want to confirm if the canonicalisation check is still required at all? If not we could remove

indices = (self.indices if self.indices[0] < self.indices[-1] else self.indices[::-1])

and simply return the representation in the order of self.indices.

— Reply to this email directly, view it on GitHub https://github.com/MDAnalysis/mdanalysis/issues/4181#issuecomment-1626900633, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACGSGB2PB7IGQK7QIQSMMO3XPEGXDANCNFSM6AAAAAAZWIHAY4 . You are receiving this because you commented.Message ID: @.***>