rdkit-repr

To compute RDKit molecular descriptors from SMILES data, run the provided CLI script on a CSV or SMI input file, specifying a descriptor preset such as Lipinski or physchem, to generate a CSV file of features for ML workflows.

You can generate multiple molecular fingerprint types from SMILES strings, including Morgan, RDKit, and MACCS fingerprints, saving the resulting feature vectors as .npy or .csv files for model training.

During featurization for QSAR modeling, the tool performs input validation and logs errors for invalid SMILES entries, ensuring that bad data does not interrupt the descriptor and fingerprint computation process.

You can calculate molecular descriptors for a single SMILES string directly from the CLI without a full dataset, allowing you to quickly generate feature vectors for individual chemical structures.

Yes, you need pandas and numpy installed alongside rdkit, as these dependencies support the input data processing and the output of numerical fingerprint arrays for machine learning pipelines.

The best way to list available RDKit descriptors and presets for feature selection is to use the CLI's built-in listing function, which quickly displays the physchem, topological, and Lipinski options for QSAR experiments.