{"title":"Revolutionizing toxicity predictions of diverse chemicals to protect human health: Comparative QSAR and q-RASAR modeling","authors":"Shubha Das, Probir Kumar Ojha","doi":"10.1016/j.toxlet.2025.06.019","DOIUrl":null,"url":null,"abstract":"<div><div>The extensive application of chemicals in the form of pesticides, cosmetics, drugs, etc., has been shown to adversely affect humans and the environment, mainly through food product residues and environmental exposure. Exposure to diverse chemicals through various routes including ingestion, inhalation, and dermal contact is associated with multiple health risks including endocrine disruption, cancer, and neurotoxicity. This study presents an advanced computational approach using quantitative structure-activity relationship (QSAR) and quantitative read-across structure-activity relationship (q-RASAR) models to predict the acute toxicity of diverse chemicals in humans, with the negative logarithm of the lowest published toxic dose (pTDLo) endpoint. We developed the first-ever predictive toxicity models combining QSAR and similarity-based read-across techniques to enhance accuracy, utilizing the TOXRIC database. The q-RASAR model outperformed traditional QSAR approaches, achieving robust statistical performance with internal validation metrics of R<sup>2</sup> = 0.710, Q<sup>2</sup> = 0.658, and external validation metrics of Q²<sub>F1</sub> = 0.812, Q²<sub>F2</sub> = 0.812, Δr<sup>2</sup><sub>m(test)</sub> = 0.087 and <span><math><mover><mrow><msubsup><mrow><mi>r</mi></mrow><mrow><mi>m</mi><mo>(</mo><mi>test</mi><mo>)</mo></mrow><mrow><mn>2</mn></mrow></msubsup></mrow><mo>̅</mo></mover></math></span> = 0.741. The model identified the key structural features, such as high coefficients and variations in similarity values among closely related compounds, the presence of carbon-carbon bonds at specific topological distances (5 and 8), and higher minimum E-state indices, all of which are linked to increased toxicity toward humans. The PLS-based q-RASAR model was further utilized to screen pesticides obtained from the pesticide properties database (PPDB) and 3660 investigational drugs from the DrugBank database for potential toxicants in humans, providing a tool to identify hazardous substances and mitigate risks. The developed models are instrumental in filling eco-toxicological data gaps and facilitating the development of novel, safe, and eco-friendly chemicals.</div></div>","PeriodicalId":23206,"journal":{"name":"Toxicology letters","volume":"411 ","pages":"Pages 16-24"},"PeriodicalIF":2.9000,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Toxicology letters","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0378427425001286","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"TOXICOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
The extensive application of chemicals in the form of pesticides, cosmetics, drugs, etc., has been shown to adversely affect humans and the environment, mainly through food product residues and environmental exposure. Exposure to diverse chemicals through various routes including ingestion, inhalation, and dermal contact is associated with multiple health risks including endocrine disruption, cancer, and neurotoxicity. This study presents an advanced computational approach using quantitative structure-activity relationship (QSAR) and quantitative read-across structure-activity relationship (q-RASAR) models to predict the acute toxicity of diverse chemicals in humans, with the negative logarithm of the lowest published toxic dose (pTDLo) endpoint. We developed the first-ever predictive toxicity models combining QSAR and similarity-based read-across techniques to enhance accuracy, utilizing the TOXRIC database. The q-RASAR model outperformed traditional QSAR approaches, achieving robust statistical performance with internal validation metrics of R2 = 0.710, Q2 = 0.658, and external validation metrics of Q²F1 = 0.812, Q²F2 = 0.812, Δr2m(test) = 0.087 and = 0.741. The model identified the key structural features, such as high coefficients and variations in similarity values among closely related compounds, the presence of carbon-carbon bonds at specific topological distances (5 and 8), and higher minimum E-state indices, all of which are linked to increased toxicity toward humans. The PLS-based q-RASAR model was further utilized to screen pesticides obtained from the pesticide properties database (PPDB) and 3660 investigational drugs from the DrugBank database for potential toxicants in humans, providing a tool to identify hazardous substances and mitigate risks. The developed models are instrumental in filling eco-toxicological data gaps and facilitating the development of novel, safe, and eco-friendly chemicals.