Toxicity estimation by chemical substructure analysis: the TOX II program

Area of Science:

• Environmental health
• Toxicology
• Computational chemistry

Background:

• Assessing chemical toxicity is crucial for environmental health and drug development.
• Limited toxicological data hinders the evaluation of novel or uncharacterized molecules.
• Predictive toxicology methods are needed to identify potential hazards early.

Purpose of the Study:

• To develop methodologies for identifying potential environmental health hazards of chemicals.
• To enable the evaluation of molecules with limited or no existing toxicological information.
• To predict the toxicity of new molecules with a reasonable degree of certainty.

Main Methods:

• Utilized MULTICASE, an artificial intelligence program, to establish structure-toxicity relationships.
• Employed TOX II, a program designed to identify toxic substructures in new molecules.
• Evaluated potential toxicity, including automatically generated metabolites, across over 70 toxicological endpoints.

Main Results:

• Successfully developed AI-driven methodologies for predicting chemical toxicity.
• Demonstrated the capability to identify toxic substructures in novel organic molecules.
• Established a framework for predicting toxicity for over 70 endpoints, including metabolites.

Conclusions:

• AI-powered tools like MULTICASE and TOX II offer reliable methods for predicting chemical toxicity.
• These predictive techniques are invaluable for assessing environmental health risks of new chemical entities.
• The developed methodologies significantly advance the field of predictive toxicology and chemical safety assessment.