Design decision: polymers and hierarchy

[avirshup]

From @jchodera on slack:

Was there any consideration of a more complex hierarchical organization aside from atoms/residues/chains? Some representation of independent molecules might be nice (since residues can be either polymeric residues within a molecule or individual molecules)

There could also be higher-order assemblies, though the "chains" concept may extend to this.

From @avirshup on slack:

I'm of 2 minds about this.

• A) Flexible hierarchies are great, because ... well, they're flexible. A scientist should have no problem coming up with a custom hierarchy to represent, say, a dendritic cellulose structure wrapped around a carbon nanotube. However, custom hierarchies mean that everyone will do it differently.
• B) PDB-style hierarchies - These are, fully, "Entity/Chain/Residue". The nice thing is that almost all computational chemists will understand this scheme, even if it's a bit of a stretch for abiological or non-polymeric components. PDBx/mmCIF has some extensions to this, although I haven't seen any software that supports it.

In MDT, I went with choice B), because I, at least, find it intuitive to access an atom by typing mymolecule.chain['A'].residue['PHE123'].atom['CA'].

In contrast, I think mdanalysis went with choice A), using abstract, nested "groups"

[davidlmobley]

There are a lot of decisions in our field which seem to originate with what part of chemistry you care about first, and a lot of them -- for biomolecular simulations -- seem to originate from people caring first about proteins.

For whatever reason I've sort of been on the "small molecule" end of things for a while and I tend not to like the entity/chain/residue models very well. Also, as I think about bigger and bigger small molecules (host-guest systems -- cucibit[n]uril as hosts, for example -- are the beginning of this) I start finding that I would like some sort of hierarchy. Obviously polymers, etc. will be a whole different level of that.

Probably we should grab someone from chemical engineering who thinks about these things and get them in on this discussion. Maybe @mrshirts will have thoughts.