Know more

Dissertation abstract:

Today, few methods allow modeling of these different types. A major problem is the heterogeneity of data. Some can be easily modeled with Gaussian Laws (transcriptomics data by DNA arrays), others are known but non-gaussian (epigenomics data from methylation profiling arrays), and still others are discrete (transcriptomic and epigenomic data by sequencing, genotyping). However, mathematical and statistical models depend strongly on the type of data; this is the reason for which the models proposed up to now have mainly been done for each type of data independently.

In this PhD study, we will propose a conjoint model for multiomics data based on the theory of copula methods. These allow the modeling of interactions between random variables of different nature and are therefore well adapted to our objectives. Several methodological challenges will be studied in order to broaden this inference method to multiomic regulation networks. The modeling proposed will be applied to biological problems studied at INRAE, and in particular those raised by the Animal Genetics Division of INRAE.

Confinanced PhD study
Start of the project: October 2022
Hosting Unit: UR 1404 MaIAGE (Mathematics and Informatics Applied to the Genome and the Environment)
INRAE Research Center: Jouy-en-Josas
University: University Paris Saclay
Doctoral School: EDMH Doctoral School
Discipline / Specialty: Mathematics of interfaces
PhD supervisors: Gildas MAZO (INRAE, MaIAGE), Florence JAFFREZIC (INRAE, GABI)
Financing : DIGIT-BIO Metaprogramme/ EDMH Doctoral School
Metaprogramme Fields of Research: field 1 (Multiscale decrypting of living functions: regulations and integration)

Contact :

• Ekaterina Tomilina (UR MaIAGE)