Exposure to chemicals triggers a series of effects at the molecular level. Regulatory pathways involved in such responses exhibit changes of levels, interactions, and feedback loops of biomolecules of different types that are active in complex networks.
Different omics techniques are essential for measuring responses in an untargeted manner on the molecular level. Importantly, a single omics technique e.g. transcriptomics, will detect biomolecules of one type and thus captures changes only for a small subset of the components of a particular pathway. Therefore, applying single omics analyses in response to a toxicant in a non-continuous design led to the identification of biomarkers for certain exposures but not to a systemic understanding of toxicity pathways or adverse outcome pathways (AOPs) in the past. Also, this incomplete representation of pathways in single omics data limits the ability to discriminate adaptive from adverse molecular responses.
A substantial improvement in detecting the pathway response to a toxicant can be achieved by using multi-omics data in a time- and concentration-resolved design, which has been recently propounded in an opinion article (Escher, Hackermüller et al. 2017). It will initially compile a collection of existing multi-omics data sets, including the integration of BASF-generated metabolomics data with published omics data sets, to:
- Evaluate various published and in house developed integrative analysis approaches for multi-omics data, in particular making use of an approach we have recently developed to integrate proteomics and transcriptomics data which combines network inference and comparative genomics
- Identify criteria for an optimal design of a multi-omics toxicity study.
Based on the lessons learnt, project representatives will perform an oral rat toxicity study focusing on direct and indirect thyroid toxicity. Biomaterials from this study will be used to generate transcriptomics, proteomics, and metabolomics data sets. Featuring a time- and concentration-resolved design including recovery after treatment this study allows investigating adaptive versus adverse effects in several dimensions. Using this multi-omics data set, the project aims to:
- Detect toxicant-triggered re-wiring of regulatory networks and changes of master-regulators, including non-coding RNAs, that link to adverse-outcomes.
- Evaluate whether time- and concentration-resolved multi-omics data enable the prediction of adversity.
- Evaluate whether a deep machine learning approach may assist in discriminating adaptive versus adverse perturbations of regulatory networks.
- Assess whether such integrative analyses enable discriminating toxicant-induced versus disease-caused, secondary perturbations.
- Determine to what extent individual omics experiments contribute to the predictive power of such an approach.
Given that most toxicogenomics studies investigate exposure to one concentration at one time point using a single omics technique, the approach significantly advances the state of the art – in the design of the animal experiment, the multi-omics perspective, and the innovative bioinformatic analyses it proposes.
The proposed project perfectly aligns with the LRI objective “Innovating Chemical Testing”. In combining an animal toxicity study with a multi-omics approach and advanced in silico methods for data interpretation, this project will contribute to “New research tools, such as […] genomics, present exciting approaches that have the potential to link information at the molecular level to health and environmental impacts”. The approach aims at refining animal experiments, in optimizing the design for a multi-omics approach, which will maximize the information gained per animal. As it aims at predicting adverse responses based on multi-omics data obtained early in an animal toxicity study, our approach may allow to reduce animal testing and thus contributes to “Towards protecting and improving animal welfare”.