Novel Quantitative Structure–Activity Relationship Tox21 Techniques for Combined Toxicity Prediction
Na Li · 2025
Humans and animals are exposed to mixtures of various environmental pollution; however, there is limited toxicity data for chemical mixtures, and the traditional methodologies for evaluating the effects of chemical mixtures including concentration addition (CA) and independent action (IA) models have been increasingly challenged and replaced. The computational approaches of quantitative structure–activity/property/toxicity relationship (QSAR/QSPR/QSTR) are already proven efficient alternatives for assessing the toxicity of chemical mixtures. In this chapter, the QSAR models for predicting endocrine-disrupting activities and acute toxicities, as well as the QSAR models based on machine-learning method, biomolecular interaction networks, toxicokinetic–toxicodynamic studies, high-throughput transcriptomics approach, and geospatial modeling approach have been reviewed. The prediction of the toxicity of chemical mixtures needs to be integrated for a comprehensive systems-level analysis to identify their toxicity effect by integrating chemical bioactivity data including chemical bioactivity, targets and pathways, gene expression, protein interactions, and the localized chemical exposure data, which will help provide a comprehensive and solid foundation for chemical mixtures prediction and analyses.