Machine learning-assisted identification of environmental pollutants by liquid chromatography coupled with high-resolution mass spectrometry

Abstract

Chemical exposure can be linked with various adverse effects, but the causal association is still poorly understood. To meet the challenge, non-target screening (NTS) based on liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) is increasingly applied to identify known and unknown chemicals with toxicological concerns present in environmental and biological samples. In this review, we highlight that the integration of predictive toxicology in NTS workflows enables large-scale screening for emerging chemical contaminants. We summarize the applications of machine learning (ML) and deep learning (DL) in toxicity prediction with a focus on biological pathway perturbation and LC-HRMS data processing, especially in peak picking and molecular structure elucidation. The substantial progress in computational approaches allows for identifying and prioritizing emerging chemical contaminants with improved accuracy, reproducibility, and efficacy. ML and DL will become next-generation informatics tools in NTS workflows to better characterize exposure to environmental pollutants.