Datasets, alphabets and models from paper 'Reverse Engineering Molecules from Fingerprints through Deterministic Enumeration and Generative Models.

Files utilized and produced within the molecule-signature project: alphabets.zip: Alphabets of molecule signatures built from MetaNetX, eMolecules and ChEMBL datasets.zip: Datasets from MetaNetX, eMolecules, DrugBank and MolForge used to build alphabets, train generative models, and evaluate methods. models.zip: PyTorch/Lightning models and SentencePiece tokenization models for decoding SMILES from ECFP. See embedded README.md files and the publication for in depth details.