The emergence of omics technologies has transformed how we study and characterise the responses of biological systems to xenobiotics. Using pathways analysis to interpret responses through bioinformatics approaches and machine learning methods to build predictive models have become new paradigms for processing complex omics data. In this study, three different machine learning approaches were proposed to investigate the relationship between chemicals and their regulatory pathways.

The first project presents a Quantitative Structure-Activity Relationships (QSAR) model that innovatively predict regulatory pathways of chemicals. The findings underscore the potential of predictive chemical-pathway associations, enhancing our understanding of chemical mechanisms of action and their side effects. While QSAR model gives satisfactory results, data limitations persist, such as the factors that may affect the pathways, which are lack of curation. Hence introducing the need of Natural Language Processing model for information extraction from omics literature. A semi-automated structured data retrieval method was developed, effectively processing 13,234 articles to extract critical omics study details and results. This approach facilitated the creation of a comprehensive dataset that could significantly advance the understanding of toxicological mechanisms and biological responses to various perturbations.

The third project provides an in-depth exploration of the factors that influence pathways and showcases a multivariate multi-omics integration to enhance chemical risk assessments. By integrating transcriptomics, proteomics, and phosphoproteomics data, the study identifies dose-dependent and temporal variances in response to chemical exposures. Advanced analytical techniques have allowed for a detailed understanding of the underlying mechanisms, showcasing the utility of omics technologies in predicting adverse outcomes.

The implications of these findings are significant, offering a more systematic approach to toxicological assessments and a better understanding of biological responses to chemicals. This research not only supports the reduction of animal testing but also enhances the development of more precise predictive models for chemical safety.