Toxicity assessment of agrochemical and pharmaceutical metabolites using artificial intelligence machine learning and computational approaches

Abstract

Pharmaceuticals and agrochemicals are the emerging pollutants (EPs) of great concern for ecosystem due to their toxicophoric moieties in chemical structure. This concept extends to metabolite toxicity, as metabolites typically formed for detoxification, sometimes more or less toxic effects as compared to xenobiotics. Individual susceptibility to toxicity hinges on the balance between a drug’s bioactivation into a toxic metabolite and its detoxification. Understanding the mechanism of toxicity can prevent metabolite-mediated toxicity with some limitations. While the toxicity profile of parent xenobiotics is established through experimental and computational techniques, the toxicity profile of metabolites procured from xenobiotics remain limited due to short instability, isolation difficulties and complex biological interactions. These challenges in in-vivo techniques highlight the value of in-silico approaches which more relevant for toxicity profiling of metabolites. The application of in-silico toxicology not only explores hazards such as organ toxicity and predicted toxicity endpoints but also forecasts biotransformation pathways. Advancements in in-silico techniques offer promising tools for assessing the toxicity of these metabolites before they reach widespread exposure. The review revisits metabolite toxicity using machine learning (ML), artificial intelligence (AI) and computational tools for metabolite formation and toxicity prediction. We highlighted the various metabolome databases useful for the data collection and metabolite analysis. Additionally, we offered some AI and ML based in-silico tools useful for the metabolite prediction and their toxicity assessment (TA).