Is there any update for this PR? I looked at the test results, and it seems this CI would work. There are two notebooks that failed to run. One seems due to an outdated import, and the other one failed because of a matplotlib import error. Can you install this package in the CI when running the tests on those examples? We can wait for #834 to be merged and run the tests again.

With the -v verbose tag, I was able to make the notebook test fails CI. After adding all the notebooks to CI, there are some bugs found.

mpnn_fingerprints notebook is now failing as it uses matplotlib, but chemprop env doesn't come with matplotlib. @shihchengli What do you think we should do? Should we add matplotlib to chemprop env? Or should we remove the plotting examples?

extra_features_from_featurizer notebook is also now failing due to cannot import name 'MoleculeFeaturizer' from 'chemprop.featurizers' error. I assumed that it intends to import SimpleMoleculeMolGraphFeaturizer, so I changed that. However, after fixing that, now the notebook fails with the following error during initialization of the MorganFingerprintMoleculeFeaturizer. @kevingreenman @KnathanM do you two know how to fix this?

mf = MorganFingerprintMoleculeFeaturizer()
morgan = mf(utils.make_mol(smis[0], keep_h=False, add_h=False))
morgan.shape, morgan
>>>
TypeError                                 Traceback (most recent call last)
Cell In[8], [line 1](vscode-notebook-cell:?execution_count=8&line=1)
----> [1](vscode-notebook-cell:?execution_count=8&line=1) mf = MorganFingerprintMoleculeFeaturizer()
      [2](vscode-notebook-cell:?execution_count=8&line=2) morgan = mf(utils.make_mol(smis[0], keep_h=False, add_h=False))
      [3](vscode-notebook-cell:?execution_count=8&line=3) morgan.shape, morgan

File <string>:6, in __init__(self, atom_featurizer, bond_featurizer, extra_atom_fdim, extra_bond_fdim, fp_size)

TypeError: MorganFingerprintMoleculeFeaturizer.__post_init__() takes 1 positional argument but 3 were given

As matplotlib is widely used in visualization, I am fine with including it as a dependency. If other devs do not want to include it as a dependency because we never use it in the other parts, we can consider removing the figures from the notebooks or installing it additionally during testing. As for the featurization issues, I suggest you wait until the PR #834 is finished, which should be able to resolve these issues.

@shihchengli I install matplotlib before the notebook test. I also talked to @KnathanM to figure out a way to fix the featurizer notebook. Could you review and approve? I will uncomment the unit tests after reviews.