The Mistra SafeChem in silico toolbox presented at the SETAC conference

The novel trustworthy and user-friendly chemical hazard prediction toolbox, developed by Mistra SafeChem, was presented by Ziye Zheng, IVL Swedish Environmental Research Institute, at the SETAC annual conference.

SETAC, Society of Environmental Toxicology and Chemistry, held its European annual meeting in Seville, Spain, in the beginning of May 2024. Ziye Zheng from IVL Swedish Environmental Research Institute was there to present the Mistra SafeChem in silico toolbox.

Machine learning prediction for chemical hazard screening has been highlighted in recent years. However, there are several problems with the available tools, such as outdated training sets, no consideration of recently published chemical records, and a lack of a graphical user interface.

67 newly trained hazard prediction models

To address these problems, the machine-learning based Mistra SafeChem in silico toolbox has been developed. It includes 67 newly trained hazard prediction models, including traditional machine-learning models with chemical descriptors and descriptor-free deep-learning models.

The models cover a wide range of chemical hazard endpoints:

• Environmental hazards (persistence, bioaccumulation, and biodegradation)
• CLP-focused endpoints (carcinogenicity, mutagenicity, and reproductive toxicity)
• Organ-specific toxicities (skin sensitization, eye irritation, and eye corrosion)
• 26 different endocrine toxicities

"A user-friendly platform for chemical risk management"

The toolbox covers a wider chemicals space, provides increased trustworthiness and gives – in most cases – three models for each endpoint. It is packaged into a Windows executable file with a user-friendly graphical interface.

– We believe our toolbox provides a user-friendly platform for chemists, toxicologists, and stakeholders for chemical risk management to predict hazards, says Ziye Zheng, who has developed the toolbox together with Ulf Norinder, Stockholm University, and Swapnil Chavan and Ian Cotgreave, RISE, within the work of Mistra SafeChem.

For more details, see the poster: A Novel Trustworthy and User-friendly Chemical Hazard Prediction Toolbox Pdf, 1.1 MB, opens in new window.