De Novo KOR Antagonists Datasets

Datasets used as input, generated or analyzed in the manuscript <strong>De Novo Design of </strong><strong>κ</strong><strong>-Opioid Receptor Antagonists Using a Generative Deep Learning Framework</strong>: 1. Pre-training Dataset #1 (4.6 million compounds from ZINC 20): zinc20_train.csv (494 MB)<br> 2. Pre-training test set (0.5 million compounds from ZINC 20): zinc20_test.csv (55 MB)<br> 3. Pre-training test scaffolds set (0.5 million compounds from ZINC 20): zinc20_test_scaffolds.csv (57 MB)<br> 4. Pre-training Dataset #2 (2,056 compounds from ChEMBL with IC50 data): chembl30_kor_inhibitors.csv (144 KB)<br> 5. 169 compounds prioritized for chemical synthesis on the basis of their similarity to JDTic's interactions with KOR: prioritized_169candidates_from1M_by_SIFt_Tc.csv (17 KB)<br> <br> All datasets are in Simplified Molecular Input Line Entry System (SMILES) format.