Improvement suggestions for chirality considerations in Arkane

[alongd]

Motivation or Problem

This issue addresses both a bug and a feature suggestion (I imagine both will be addressed in a single future PR).

The Bug

As noted by @goldmanm, our method for determining the number of optical isomers in Arkane (based on rmgpy.qm.qmdata.QMData functionalities, and ostensibly on S. Patchkovskii's code) is buggy.

The Feature

Currently Arkane treats the number of optical isomers as a boolean (more correctly, either 1 or 2 for no chirality or any number of chiral centers, respectively.) I think that when Arkane was originally written, it dealt with small species, and not many chiral centers. However, dealing with larger molecules (e.g., drug-related) introduces many chiral centers, which should be taken into account in the partition function. Moreover, users may want to consider nitrogen inversion modes as potential chiral centers (if connected to three different groups, considering the lone pair as the forth unique group), as well as cis/trans chiralities around a double bond, depending on the temperature.

Desired Solution

The suggestions are to:
(a) allow the user to specify which chiralities they care about (C/N/D, standing for carbon, nitrogen, double bond); 'none' could also be specified if the mix has only one diastereomer.
(b) automatically determine the correct number of chiral centers accordingly (by type) if they were not specified
(c) consider the centers of interest in the partition function for better entropy/rates calculations.

Additional Context

ARC already has functionalities that could be used for this end. Unfortunately, since RMG/Arkane are dependencies of ARC, we'll need to duplicate these methods instead of import them.

Below are some species compiled by @goldmanm with a comparison between the two approaches (all species in this list have no chiral centers but are predicted to have chirality according the the present method):

label	smiles	correct number of chiral centers	Arkane's number of chiral centers*	number of chiral centers determined by the new approach
aR	CC(C)[CH]O	0	1	0
b-bscission From Alkoxy	C=O.C[C](C)OO	0	1	0
bAlkoxyIsom	CC(C)(C[O])OO	0	1	0
bH2OForm	CC(C)([CH]O)OO	0	1	0
bQOOH[O]	CC(C)(C[O])OO	0	1	0
bQOOHa	CC(C)([CH]O)OO	0	1	0
bR	C[C](C)CO	0	1	0
bRO2	CC(C)(CO)O[O]	0	1	0
galkene	C=C(C)CO	0	1	0
ibut_peroxyacid	CC(C)C(=O)OO	0	1	0
ibutaldehyde	CC(C)C=O	0	1	0
ipropylOOH	C[CH]COO	0	1	0
monosub_c4ether	OCC1COC1	0	1	0
propene3ol	C=CCO	0	1	0

Note: Arkane's number of chiral centers is the number of optical isomers - 1

(while it's easy to come up with a function that always spits out 0s... it does actually work for other cases too...)

[alongd]

I'm not sure whether or not to implement this approach: It is based on RDKit, so I don't expect it to work for TSs.

[github-actions[bot]]

This issue is being automatically marked as stale because it has not received any interaction in the last 90 days. Please leave a comment if this is still a relevant issue, otherwise it will automatically be closed in 30 days.