testBondDoubleStereoProblem1 test fails in Python 3.9.8 and rdkit versions >=2021.9.1b1

[djm-healx]

The assertion in the following statement fails when running unittest in Python 3.9.8, using rdkit version 2022.3.1: https://github.com/chembl/ChEMBL_Structure_Pipeline/blob/8d599ad389a458c002be9fd8353a91ebaf370743/chembl_structure_pipeline/test/test_standardizer.py#L1655

Which raises:

FAIL: testDoubleBondStereoProblem1 (test.test_standardizer.TestCase)
a problem found in testing
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/danielmason/Healx/repositories/ChEMBL_Structure_Pipeline/chembl_structure_pipeline/test/test_standardizer.py", line 1655, in testDoubleBondStereoProblem1
    self.assertNotEqual(omolb.find('13 12  1  6'), -1)
AssertionError: -1 == -1

This was picked up in #38 when running python -m unittest discover chembl_structure_pipeline to test after creating the PR.

Is this expected behaviour, or an unexpected failure?

[djm-healx]

The tests pass using rdkit-pypi until 2021.9.1b1, after which the newer versions of rdkit cause this test to fail.

[eloyfelix]

thanks for reporting! I've just confirmed the bug and we'll need to take a look at it.

[eloyfelix]

sorry for the long silence. We had lots of work related to data lately. RDKit 2022.09.01 passes again the tests. We'll release a new version of the standardiser soon (will be included in pypi) and it will be used in ChEMBL after we release the next version, ChEMBL32