chain-level zone atomspec not yet implemented

[Elaine Meng]

“/<“ doesn’t promote to chain… gives the same set of atoms as “@<“

Example:
open 1zik
show
la :69 res
sel :69 /<5

Conrad:
The chain zone atomspec was not implemented because (in the implementation) the Chains class is completely different than Residues and Structures. We can convert between Atoms and Residues with atoms.unique_residues and residues.atoms. But there is no equivalent atoms.unique_chains or chains.atoms.

Elaine:
One contributing factor might be that there could be differing definitions of chain: same molecule/biopolymer (based on covalent bonding) or everything with the same chain ID. I guess I’d expect the latter for zone atomspec since the “/“ symbol is being used, but for selection up-down arrow cascades it’s nice to have both levels like we have in Chimera.

Eric:
That’s probably the sticking point — “chains” only cover the polymeric atoms, not other atoms with that chain ID.

Eric (earlier message):
chains<->atoms is just a “two level” conversion, i.e. chains.existing_residues.atoms and atoms.unique_residues.unique_chains.

[Conrad Huang]

Using Eric's two-level conversion for chains<->atoms, the chain zone atomspec is implemented in d06c5bea2.

[Elaine Meng]

As Eric hinted at, this approach only includes polymers. My 1zik example should select everything with chain ID A (peptide + solvent) but it only gets the peptide.

In other words, I expected "select :69 /<5" to give the same results as "select /a" since chain A has atoms within 5 angstroms of residue 69. Seems to be worth discussing, at least.

[Conrad Huang]

Fixed in 332a29749. Chain zone atomspecs now use chain ids (instead of polymers) for selecting atoms.