Discovery of Selective Nrf2 Activators from Natural Products: A Computational Screening Approach to Minimize Off-Target Effects on PXR and CYP2D6

This study provides a comprehensive resource for large-scale analysis of natural product selectivity toward Nrf2-related targets. It includes molecular docking results, physicochemical properties, and cheminformatic annotations for 628,898 natural products screened against three key proteins: KEAP1 (Nrf2 regulator), PXR (Pregnane X Receptor), and CYP2D6 (Cytochrome P450 2D6). Each compound entry contains KEAP1, PXR, and CYP2D6 docking scores (karma_score, karma_score_ff, and karma_score_aligned), together with calculated selectivity indices (NSI, SI, and CSS), molecular descriptors (e.g., MW, LogP, TPSA, Fsp³, aromatic ring count), and drug-likeness metrics (QED, Lipinski’s violations). Chemical classifications from the COCONUT database are also included, covering hierarchical taxonomies (superclass, class, subclass) and NP Classifier biosynthetic pathway annotations. The dataset thus integrates structural, functional, and toxicological dimensions into a unified framework suitable for data mining, model development, and cross-study benchmarking. The code and intermediate data (version 1)uploaded here is provided only as an illustrative example; it does not represent the final production version used in the study, and further revisions were made during the paper’s review process. The full COCONUT molecular docking score data was uploaded in version 2.