LibINVENT: Reaction-based Generative Scaffold Decoration for in Silico Library Design

Training datasets used for LibINVENT publication https://doi.org/10.1021/acs.jcim.1c00469. The datasets used for training of the prior: purged_chembl_sliced.smi.gz: The CHEMBl 27 compounds, filtered according to the rules described in the manuscript and sliced according to the reaction rules. chembl_train.smi.gz: The purged, sliced dataset used for model training. The DRD2 compounds are removed as described in the manuscript.