How to compute RDKit molecular features in Mathematica…

First, initialize your code by creating an external python evaluation session which sets up a temporary virtual environment for RDKit (and associated dependencies), and then begin by importing the relevant RDKit libraries:

session = ResourceFunction["StartPythonSession"][{"rdkit"}]
ExternalEvaluate[session, "from rdkit.Chem import MolFromSmiles, Descriptors"]

12ss2xaemp3q8

Second, define a function (features) which will run in that session:

features = ExternalFunction[session, 
   "def features(smiles):
      m = MolFromSmiles(smiles)
      vals = Descriptors.CalcMolDescriptors(m)
      return vals"]

0lat68wk70a34

You may now use the defined ExternalFunction like any other Mathematica function:

features["CCN(CC)C(=O)[C@H]1CN([C@@H]2CC3=CNC4=CC=CC(=C34)C2=C1)C"]

0ic5gwwh3h1x5

Be a good citizen and clean up your session when you are finished:

DeleteObject[session]

Parerga and Paralipomena

“Under the hood”, Mathematica actually uses RDKit to compute many of the various MoleculeValue properties; however these do not include all of the descriptors, nor do the names match.
If your goal is to compute binary fingerprint vectors, it is more elegant to use the MoleculeFingerprints paclet instead

ToJekyll["How to compute RDKit molecular features in Mathematica", "chemistry cheminformatics mathematica"]