Meta-analyzed datasets for toxicity classification of metal oxide nanoparticles

This dataset package contains four curated datasets (MeOX I, MeOX II, MeOX IIIA, and MeOX IIIB) that were originally compiled and published through the S2NANO portal (Safe and Sustainable Nanotechnology, portal.s2nano.org). The datasets compile physicochemical, quantum-mechanical, and in vitro toxicological properties of 26 types of metal oxide nanoparticles extracted from 216 published studies. MeOX I: Raw compiled dataset with missing values in key physicochemical attributes (core size, hydrodynamic size, surface charge, specific surface area). Mean substitution was applied during modeling. MeOX II: Dataset with missing physicochemical data filled via manufacturer specifications and estimations, enhancing data completeness and quality. MeOX IIIA: Subset of MeOX II, selecting the top 50% of samples based on a physicochemical (PChem) data quality score to emphasize high-quality data. MeOX IIIB: Further refined subset of MeOX II, containing the top 20% highest-scored samples for the highest data quality assurance. Each dataset is labeled for binary classification (Toxic vs. Nontoxic) based on 50% cell viability thresholds under various experimental conditions. These datasets support nano-SAR model development and facilitate research into structure-activity relationships for engineered nanomaterials.