Mots-Clés

Artificial Intelligence transcriptomics imaging toxicology chemical risk assessment

Description

Towards the objective to reduce animals testing for the safety assessment of chemicals, it has been recently accepted that new alternatives methodologies (NAM) such as in vitro screening, omics data, systems biology and their integration through computational models could be relevant in extrapolation of in vitro experimental data to predict in vivo phenomena and so regulatory decision making.

In this context, the objective of the project will consist to link transcriptomics outcomes with phenotypic data (i.e. cell imaging) for a large set of chemicals and to develop predictive models based on machine learning and artificial intelligence in the aim to decipher a chemical-genes signature associated to cell perturbation and further combined with exposure estimates can lead to adverse outcomes. The results of such analysis will be in a second step linked to in vivo animal studies for a set of chemicals in order to suggest mechanistic relationships to in vivo toxicity. The ultimate goal of this Ph.D project is to propose a new alternative approach based on artificial intelligence able to predict the risk of toxicity for new substances on which human can be exposed.

The project will be performed at the CMPLI team (U1133, Inserm), member of the biological and functional adaptive unit (CNRS UMR 8251), Université de Paris under the supervision of Prof Olivier Taboureau and in close collaboration with Dr. David Rouquié, toxicologist at Bayer Cropscience and Affiliate Chair of the 3IA institute Côte d'Azur.

It is required for the candidate to have some skills in bioinformatics, omics data analysis, programming (python, R) and machine learning (deep learning). Knowledge in toxicology, transcriptomics and high content imaging are suitable but not mandatory.