Marie Perrot-Dockes

Regularization tools for multivariate analysis: application to multi-omics (2016-2019)

Marie’s PhD was co-supervised with Céline Lévy-Leduc (50%/50%) and Laure Sansonnet, from 2016 to 2019.

Summary

This thesis develops regularized methods for variable selection in high-dimensional multivariate linear models, where the response is a vector of correlated outcomes rather than a single variable. It proposes procedures that jointly estimate the regression coefficients and the covariance structure of the residuals, improving selection accuracy when responses are strongly dependent, and establishes theoretical guarantees for the resulting estimators. The methodology is implemented in the MultiVarSel R package and applied to metabolomics data (LC-MS), where it is used to identify which molecular features are jointly associated with experimental conditions while accounting for the complex correlation structure between metabolites.

PhD manucript

2019

  1. PhD thesis
    Méthodes régularisées pour l’analyse de données multivariées en grande dimension: théorie et applications
    Marie Perrot-Dockès
    Université Paris-Saclay, 2019

Journal papers

2019

  1. Cell
    A Quantitative Multivariate Model of Human Dendritic Cell-T Helper Cell Communication
    Maximilien Grandclaudon, Marie Perrot-Dockès, Coline Trichot, and 9 more authors
    Cell, 2019

2018

  1. SAGMB
    A multivariate variable selection approach for analyzing LC-MS metabolomics data
    M. Perrot, C. Lévy-Leduc, Julien Chiquet, and 5 more authors
    SAGMB, 2018
  2. JMVA
    Variable selection in multivariate linear models with high-dimensional covariance matrix estimation
    M. Perrot, C. Lévy-Leduc, L. Sansonnet, and 1 more author
    J. Multivar. Anal., 2018

Conferences

2018

  1. Multivariate statistical modelling for QTL detection and marker selection in a bi-parental grapevine population
    Charlotte Brault, Marie Perrot-Dockès, Agnes Doligez, and 3 more authors
    In 17. Meeting of the EUCARPIA Section Biometrics in Plant Breeding, 2018

2018

  1. Sélection de variables dans le modèle linéaire multivarié en grande dimension avec prise en compte de la dépendance
    M. Perrot, C. Lévy-Leduc, J. Chiquet, and 1 more author
    In actes des 50\ieme journées françaises de statistique, Saclay, 2018

2017

  1. Modèle linéaire multivarié parcimonieux avec estimation de covariance : une application à des données de métabolomique
    M. Perrot-Dockes, C. Lévy-Leduc, J. Chiquet, and 1 more author
    In actes des 49\ieme journées françaises de statistique, Avignon, 2017