Molfeat

If you're doing molecular machine learning, this gives you 100+ ways to turn SMILES strings into feature vectors. It handles everything from classic fingerprints like ECFP and MACCS to pretrained transformer embeddings like ChemBERTa. The three-tier structure is smart: calculators for single molecules, transformers for batches with scikit-learn compatibility, and pretrained models for deep learning embeddings. Parallel processing works out of the box, and you can save configurations as YAML for reproducibility. The real win is not having to cobble together RDKit, DGL, and transformer libraries yourself. Whether you're building QSAR models or screening a million compounds, it consolidates the featurization step into one consistent API.