From the README:

TO DO
Add code to remove stereo in Utils.standardize_chemical, update tests accordingly.

Stereochemistry is not handled and raises a rpreactor.Utils.ChemConversionError and a traceback.

TODO:

• nicer "warning" message
• add a test with stereo
• implement the stereochemistry

Example:

python -m rpreactor.cli --with_hs true --match_timeout 1 --fire_timeout 1 --with_stereo true  inline --inchi "InChI=1S/C12H22O11/c13-1-3-5(15)6(16)9(19)12(22-3)23-10-4(2-14)21-11(20)8(18)7(10)17/h3-20H,1-2H2/t3-,4-,5+,6+,7-,8-,9-,10-,11?,12+/m1/s1" --rsmarts "([#6@&v4:1](-[#8&v2:2])(-[#1&v1:3])(-[#6&v4:4])-[#6@@&v4:5](-[#8&v2:6]-[#6@@&v4:7]1(-[#1&v1:8])-[#6@@&v4:9](-[#8&v2:10]-[#1&v1:11])(-[#1&v1:12])-[#6@&v4:13](-[#8&v2:14])(-[#1&v1:15])-[#6&v4:16]-[#8&v2:17]-[#6@@&v4:18]-1(-[#6&v4:19](-[#8&v2:20])(-[#1&v1:21])-[#1&v1:22])-[#1&v1:23])(-[#1&v1:24])-[#8&v2:25]-[#6&v4:26])>>([#6@&v4:1](-[#8&v2:2])(-[#1&v1:3])(-[#6&v4:4])-[#6@@&v4:5](-[#8&v2:6]-[#1&v1])(-[#1&v1:24])-[#8&v2:25]-[#6&v4:26].[#6@@&v4:13]1(-[#8&v2:14])(-[#1&v1:15])-[#6@&v4:9](-[#8&v2:10]-[#1&v1:11])(-[#1&v1:12])-[#6@@&v4:7](-[#8&v2]-[#1&v1])(-[#1&v1:8])-[#6@&v4:18](-[#6&v4:19](-[#8&v2:20])(-[#1&v1:21])-[#1&v1:22])(-[#1&v1:23])-[#8&v2:17]-[#6&v4:16]-1)" 
09/09/2020 15:28:33 -- CRITICAL -- Stereo is not handled at the time being.
Traceback (most recent call last):
  File "/workdir/rpreactor/cli.py", line 251, in compute
    rd_mol = standardize_chemical(rd_mol, with_hs=self._with_hs, with_stereo=self._with_stereo, heavy=True)
  File "/workdir/rpreactor/Utils.py", line 77, in standardize_chemical
    raise ChemConversionError("Stereo is not handled at the time being.")
rpreactor.Utils.ChemConversionError: CHEM-CONVERSION-ERROR: Stereo is not handled at the time being.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/opt/conda/envs/dev_rpreactor/lib/python3.6/runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "/opt/conda/envs/dev_rpreactor/lib/python3.6/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "/workdir/rpreactor/cli.py", line 448, in <module>
    __cli()
  File "/workdir/rpreactor/cli.py", line 444, in __cli
    args.func(args)
  File "/workdir/rpreactor/cli.py", line 352, in inline_mode
    r.compute()
  File "/workdir/rpreactor/cli.py", line 253, in compute
    raise ChemConversionError(e) from e
rpreactor.Utils.ChemConversionError: CHEM-CONVERSION-ERROR: CHEM-CONVERSION-ERROR: Stereo is not handled at the time being.

From the README:

Buggy tales
20190110.01 -- Timeout issue

• Checking results: timeout neither reached... Maybe fine but strange.
• Checking results: bunch of error messages mentioning unknown "pool" variable (e.g. "match_error": "name 'pool' is not defined")
• Bug: typo led to wrongly handling Pool of process
• Fix: commit ca9ff17 @ rule_fire
• Imply: rerun jobs impacted (DONE, all jobs have been reran)

@tduigou, it's not clear to me if it has been fixed and the text above is merely a backlock; should we close the issue?

From the README:

Buggy tales
20190110.02 -- Empty InChI for products

• Checking results: some inchi string are empty, e.g.: rule (radius 0) 8a4666fc9dab applied on MNXM366
  • 8a4666fc9dab: ([#6&v4:1]1:[#6&v4:2]:[#6&v4:3]:[#6&v4:4]:[#7&v3:5]:[#6&v4:6]:1)>>([#6&v4:1](-[#6&v4:2]-[#6&v4:3]:[#6&v4:4]-[#7&v3:5]-[#1&v1])(-[#6&v4:6](-[#7&v3](-[#1&v1])-[#1&v1])-[#1&v1])-[#1&v1])
  • MNXM366: [H][O][c]1[c]([C]([H])([H])[H])[n][c]([H])[c]([C]([H])([H])[O][P](=[O])([O][H])[O][H])[c]1[C]([H])([H])[N]([H])[H]
• Reason: sanitization issues of products (see below) leading to empty InChI.

[16:06:30] non-ring atom 2 marked aromatic
10/01/2019 16:06:30 -- WARNING -- Partial sanization only
[16:06:30] ERROR: Unrecognized bond type: 0
[16:06:30] non-ring atom 2 marked aromatic
10/01/2019 16:06:30 -- WARNING -- Partial sanization only
[16:06:30] ERROR: Unrecognized bond type: 0

• Reason: probably due to rule miss-encoding. Below are rule known to raise with bug (checked for radius 1, 3, 5, 7, 9, 10):

Count Rule_hash
15 24f4a1100077
57 300bddc687b7
15 88eee4c1d443
57 e9f88e91236e

• Consequences: probable wrong transformations.
• Workaround: reject all results from a couple substrate-rule as soon as an "empty" InChI is produced
• Fix: commit e980692 @ rule_fire
• Imply: take care of this situation in results already generated / rerun code and make new results (DONE: new results generated)

From the README:

Buggy tales
20190110.03 -- Cannot convert back InChI to RDKit mol

• Rule (radius 0) 3012ddcc4356 applied on MNXM2183
  • 3012ddcc4356: ([#6&v4:1](-[#7&v3:2]:[#6&v4:3](:[#7&v3:4]):[#6&v4:5])-[#8&v2:6])>>([#6&v4:1]=[#8&v2].[#7&v3:2]:[#6&v4:3](:[#7&v3:4]-[#1&v1])-[#6&v4:5].[#8&v2:6]-[#1&v1])

• NXM2183: [H][N]=[c]1[n][c]([O][H])[c]2[n][c]([H])[n]([C]3([H])[O][C]([H])([C]([H])([H])[O][H])[C]([H])([O][P](=[O])([O][H])[O][H])[C]3([H])[O][H])[c]2[n]1[H]

• Reason: issue with valencies, aromatic cycle and nitrogen.
• Fix: none at the moment, but should probably be handled at the result parsing step (i.e. after rule firing)

From the README:

Buggy tales
20190112.01 -- The unkillable Pool

• Execution hang (sometimes) when renewing the Pool (done after every time out)
• Due to timed out job lock that cannot be acquired (deadlock)
• Issue do not happen every time
• 100% chance of hang when the time out == 0. Hypothesis: the job was not fully initialized, cannot acquire a lock that have not been set up
• But hang can also happen if more "reasonable" timeout condition (e.g. 5 seconds). Hypothesis: the job have been killed while it was actually ending, something went wrong.
• Straightforward fix for all cases: force to release the job lock each time. Commit 52466a5 @ rule_file

From the README.

TD201904.01 -- stereochemistry assignment from MolToSmiles(sanitize=True) is not the same as MolToSmiles + SanitizeMol

Source (april 2019): rdkit/rdkit#2361

In the second case the stereo assignment is not made:

In [44]: m = AllChem.MolFromSmiles('[O-][n+]1onc2cc(/C=C/c3ccc(Cl)cc3)ccc21', sanitize=True)

In [45]: AllChem.MolToInchiKey(m)
Out[45]: 'AALOGNDNCMFSSI-OWOJBTEDSA-N'

In [46]: m = AllChem.MolFromSmiles('[O-][n+]1onc2cc(/C=C/c3ccc(Cl)cc3)ccc21', sanitize=False)

In [47]: AllChem.SanitizeMol(m, Chem.rdmolops.SanitizeFlags.SANITIZE_ALL)
Out[47]: rdkit.Chem.rdmolops.SanitizeFlags.SANITIZE_NONE

In [48]: AllChem.MolToInchiKey(m)
Out[48]: 'AALOGNDNCMFSSI-UHFFFAOYSA-N'

For the case of calling SanitizeMol after MolFromSmiles you can
force rdkit to calculate the correct InChI key by calling
AllChem.AssignStereochemistry(m1, cleanIt=True, force=True)
before calculating the key.