Molecular datasets for fingerprint benchmarking

Two sets of compounds used to benchmark various molecular fingerprint and molecular similarity methods.compounds_ms2structures.csv Contains 37811 different compounds for which also public MS/MS spectra exist (spectra not included here).The data contains the SMILES, as well as inchikey, mass, classyfire class, subclass and superclass, formula, and for many entries also the npc class, pathway and superclass. biostructures_combined.csvThis is a set of 730464 unique organic compounds and contains the SMILES. Most of this compounds are taken from the biostructures dataset used in Kreschmer et al. (2025), https://doi.org/10.1038/s41467-024-55462-w. In addition, we merged all missing compounds from the compounds_ms2structures dataset.