Mots-Clés
Methylation
Algorithm
R
UNIX
statistic
bioinformatics
EWAS
Description
Background
DNA methylation measured using Illumina Infinium Methylation EPIC arrays is widely used in epigenome-wide association studies (EWAS). The transition from the Methylation EPIC v1.0 (EPICv1) to EPIC v2.0 (EPICv2) array increases genomic coverage but introduces differences in probe content and annotation. Although the two arrays share many CpG sites, some probes have been added, removed, or modified. Despite high overall concordance, agreement can vary across individual CpG sites, potentially affecting downstream analyses and biological conclusions. This project aims to evaluate these differences and develop an appropriate strategy for harmonizing and integrating EPICv1 and EPICv2 datasets, with particular attention to their impact on downstream EWAS results.
E3N Generation Cohort
The E3N-Generations cohort (Epidemiological Study of Women in the National Education System) is a family cohort study based on E3N, a long-standing cohort of approximately 100,000 female volunteers who are members of the MGEN (Mutuelle Générale de l’Education Nationale). Data regarding the lifestyles and health status (specifically cancer and chronic diseases) of the E3N women have been collected via self-administered questionnaires every two to three years since 1990. To address specific research objectives, dedicated questionnaires are regularly administered to subgroups within the cohort. Recruitment is currently underway for the children and fathers of the women originally enrolled in 1990. The team’s research focuses on the role of environmental, social, and genetic factors in the development of specific diseases. For the genetic part, groups work on GWAS, methylation, and metabolomic data while proteomic data is on the process of collection.
Objectives of the internship
An important scientific question in the E3N-Generations cohort is how lifestyle and environmental factors influence DNA methylation, and how these epigenetic changes may contribute to disease development. In this internship, the student will investigate associations between DNA methylation and smoking, through epigenome-wide association studies (EWAS). Smoking is a particularly relevant example because it is known to influence DNA methylation at specific CpG sites. By comparing EWAS results obtained using different harmonization and integration strategies, the student will assess whether combining EPICv1 and EPICv2 data affects the identification and reproducibility of methylation sites associated with smoking. This work will help establish reliable strategies for integrating methylation datasets and improve the robustness of future epigenetic epidemiological studies.
The project will address four main questions:
- How different are EPICv1 and EPICv2 methylation measurements?
- Can information missing from one array version be reliably inferred from the other?
- What is the most appropriate strategy for pooling EPICv1 and EPICv2 datasets?
- How does the choice of harmonization/pooling strategy affect EWAS results?
Available data
The project will use DNA methylation data generated from blood samples using two versions of the Illumina MethylationEPIC array:
- approximately 700 EPICv1 samples
- approximately 280 EPICv2 samples,
- 8 biological samples measured on both EPICv1 and EPICv2, providing paired data for direct within-individual comparison.
All participants are female controls, with information available on age, smoking, body mass index (BMI) and alcohol consumption.
Mission
The student will perform quality control and exploratory analyses to identify differences between the two platforms, including the assessment of CpG coverage, methylation concordance and potential technical effects.
The performance of selected strategies will be evaluated through EWAS of lifestyle-related phenotypes, particularly smoking, to assess their influence on association estimates and the reproducibility of biological findings.
Expected outcomes
The internship is expected to provide a better understanding of the methodological challenges associated with combining EPICv1 and EPICv2 data, identify appropriate integration strategies for future EWAS, and deliver a reproducible analytical workflow implemented in R/Bioconductor.
Candidate’s profile
Master (M2) or equivalent in bioinformatics or related fields. Strong interest in genetics and statistics. Knowledge in R programming language and UNIX environment. Knowledge of working with standard scientific repositories such as GitHub would be appreciated. The internship is in English language.
Contact
The internship will be supervised by Dr. Yazdan Asgari and Dr. Pierre-Emmanuel Sugier. Please send your CV and cover letter to Yazdan Asgari, (yazdan.asgari@inserm.fr) (Inserm U1018, CESP ‐Centre de Recherche en Épidémiologie et Santé des Populations, Villejuif).
Location
The intern will be located at the Center for Research in Epidemiology and Population Health (CESP) at Paul Brousse Hospital (Villejuif) (http://cesp.inserm.fr/).
Start date
February 2027
Reference
- Kaur et al., Comprehensive evaluation of the Infinium human MethylationEPIC v2 BeadChip, Epigenetics Communications, 3:6, 2023
- Luccier et al., Technical variability across the 450K, EPICv1, and EPICv2 DNA methylation arrays: lessons learned for clinical and longitudinal studies, Clinical Epigenetics, 16:166, 2024
- Zhuang et al., Accounting for differences between Infinium MethylationEPIC v2 and v1 in DNA methylation–based tools, Life Science Alliance, 8(9), e202403155, 2025