NewVersion 2.4.0 of the derived dataset from the USDA Dr. Duke Phytochemicals Database has been uploaded to my GitHub- and Huggingface-repos.

What the dataset includes: 76,907 records on plant compounds from 2,313 plant species, converted from the original Dr. Duke database into a structured flat file format for ML workflows.

Fields: Compound\_Name, Plant\_Species, Plant\_Part, Chemical\_Activity, PubChem\_CID, SMILES, molecular\_formula, compound\_type, number\_of\_patents\_since\_2020, method\_for\_determining\_number\_of\_patents, ClinicalTrials.gov\_flag, iupac\_verified, inchi\_key, partner\_CID, method\_for\_partner\_mapping.

What has changed in v2.4.0:

1,534 previously zero-CID records now have verified PubChem CIDs. These were resolved through a systematic IUPAC name search against PubChem REST.…