Master’s Internship (MSc – M2)
Multivariate Change Point Detection and Curve Clustering for Metabolite Kinetics Analysis
Keywords: Multivariate time series, change point detection, curve clustering, metabolomics, statistical learning
Context
This internship is part of the HepatoTwin research project (2024–2028), coordinated by INRAE and conducted in collaboration with the TOXALIM research unit, a leading laboratory in food toxicology. The HepatoTwin project aims to investigate the impact of food contaminants and dietary imbalances on liver metabolism using advanced experimental and computational approaches.
Within this project, high-frequency multivariate time series data are collected from human hepatic cell cultures, monitoring the real-time dynamics of dozens to hundreds of metabolites. Cells are exposed to different stress conditions mimicking food contaminant exposure or unbalanced diets, potentially inducing either abrupt metabolic disruptions or more gradual changes in metabolic dynamics.
Internship Objectives
The goal of this research internship is to develop statistical learning methods for detecting and characterizing metabolic perturbations. Two main scenarios will be addressed:
- Abrupt change detection, corresponding to sharp metabolic disruptions
- Pattern changes, characterized by modifications in the dynamic response profiles over time
For the first scenario, the focus will be on detecting global multivariate change points, accounting jointly for all metabolites, and on identifying groups of metabolites with similar response times or kinetic profiles. This naturally leads to multivariate methods and curve clustering approaches.
Scientific Environment and Supervision
The intern will benefit from close supervision by researchers with complementary expertise:
- Guillem Rigaill – expert in univariate and multivariate change point detection methods and algorithms
- Arthur Leroy – specialist in Gaussian processes, with applications to curve clustering
- Julien Chiquet – expert in statistical learning, regularization methods, and multivariate analysis
Methodological developments will be carried out in Python and/or R. Prototype datasets are already available to support method development and validation.
Working Conditions
- Host laboratory: UMR MIA Paris-Saclay
- Location: Agro Paris-Saclay Campus, France
- Institution: INRAE
- Duration: Master’s level internship (M2)
- Compensation: INRAE internship allowance (€4.35/hour)
- Benefits: Access to public transportation reimbursement and teleworking options
- Research environment: Access to data and scientific ecosystem of the HepatoTwin project
Contacts
- Julien Chiquet – Research Director (INRAE): julien.chiquet@inrae.fr
- Guillem Rigaill – Research Director (INRAE): guillem.rigaill@inrae.fr
- Arthur Leroy – Research Scientist (INRAE): arthur.leroy@inrae.fr