Computational identification of preservatives with potential neuronal cytotoxicity.

摘要: Abstract Preservatives play a vital role in cosmetics by preventing microbiological contamination for keeping products safe to use. However, few commonly used preservatives have been suggested be neurotoxic. Cytotoxicity neuronal cells is as the first-tier assay assessing chemical-induced neurotoxicity. Given time and resources required chemical screening, computational methods are attractive alternatives over experimental approaches prioritizing chemicals prior further evaluations. In this study, we developed Quantitative Structure-Activity Relationships (QSAR) model identification of potential neurotoxicants. A set 681 was utilized construct robust prediction using oversampling Random Forest algorithms. Within defined applicability domain, independent test on 452 showed high accuracy 87.7%. The application 157 identified 15 potentially toxic cells. Three them were validated vitro experiments. results that experiments desirable neurotoxicity with cytotoxicity.