Improve chembl drugs dataset

[hadim]

By giving more columns. We can simply backport the script made on medchem at https://github.com/datamol-io/medchem/blob/9ee009a5495d0146ec46a804fa458944711d9e28/notebooks/Get_ChEMBL_Approved_Drugs.ipynb

[hadim]

In datamol we provide a dataset of all approved drugs in ChEMBL: dm.data.chembl_drugs(). But that dataset contains only one single column: smiles.

It would be nice to have the same dataset but with some "metadata" columns such as ChEMBL ID, date of approval, etc.

For medchem, I created a small script that curate all approved drugs from ChEMBL and keep a few metadata columns.

• The script in the medchem repo: https://github.com/datamol-io/medchem/blob/e03ab8087592754645ffe12986a98b344f594547/notebooks/Get_ChEMBL_Approved_Drugs.ipynb
• The dataset curated from medchem: https://github.com/datamol-io/medchem/blob/e03ab8087592754645ffe12986a98b344f594547/docs/tutorials/data/chembl_approved_drugs.parquet

This dataset is used in the "Basic Concept" tutorial of medchem at https://medchem-docs.datamol.io/stable/tutorials/Basic_Concepts.html

---

The task:

• datamol repo:
  • add chembl_approved_drugs.parquet in datamol/data/
  • modify dm.data.chembl_drugs() to leverage it + adapt the docstring to explain how it has been generated
  • modify the unit tests (or add one if it does not exist)
  • copy the notebook from the medchem repo into a new folder in datamol: ./notebooks/

[hadim]

@stwhitfield let me know if you have any questions!