This page presents the course Statistics for Genomics co-organized by Chloé-Agathe Azencott, Florian Massip and Marie Verbanck in the Science and Technology for Health (ST4Health) MSc degree at Paris Sciences et Lettres. In 2025 and 2026, Katia Antonenko was a TA for the hands-on sessions.

This course is an 18 hour optional course in the second year of the master's program.

Course presentation

Over the last fifteen years, genomics have been transformed by high-throughput sequencing technologies, which generate large amounts of complex data that are difficult to interpret. In particular, genomic datasets often comprise a relatively small number of samples described by a large number of features, a statistically challenging setting requiring specific approaches, collectively known as high-dimensional statistics. In this setting, variable selection, clustering and model validation are known to be particularly difficult. This course will present different approaches that have been developed to overcome these difficulties. To this end, we will develop in detail a few classical and successful examples to illustrate both the power and limitations of statistics and machine learning for genomic analyses.

The course combines lectures and hands-on sessions in Python or R.

Learning objectives

  • Know the challenges of high-dimensionality for statistics (e.g. multiple hypotheses, curse of dimensionality);
  • Understand how the particularities of genomics data can impact the application of classical statistical and machine learning techniques;
  • Run statistical analyses that are standard in genomics (e.g. GWAS, differential expression analysis, matrix factorization, hierarchical or Louvain clustering);
  • Identify, for a given application case, what will be the limitations of generic statistical and machine learning techniques, and which of the techniques presented in the course (if any) can apply;
  • Determine whether a methodological paper describing a statistical method for genomics data could be relevant for a specific application case.

Prerequisites

  • Basics in statistics and machine learning
  • Linear algebra & calculus: vectors, matrices, linear equations, SVD
  • Basics in genetics

Evaluation

The final grade is based on a written quiz (30%), an oral evaluation (60%) and course engagement (10%).

The oral evaluation consists in a presentation of a scientific publication, chosen at the beginning of the course among those provided by the instructors, in small groups of 2 or 3 students. All students attend all presentation.

Schedule (2026)

  • Nov. 3 afternoon : Course presentation and Introduction to high-dimensional statistics: lecture by Chloé Azencott, hands-on session by Katia Antonenko|https://ekaantonenko.github.io/] [slides (pdf)] [hands-on (zip)]
  • Nov. 5 afternoon: Gene expression analysis: lecture by Florian Massip, hands-on session by Marie Verbanck and Katia Antonenko as TAs
  • Nov. 17 afternoon: Genome-wide association studies: lecture by Chloé Azencott and Mendelian randomization: lecture by Marie Verbanck
  • Dec. 1st afternoon: Genome-wide association studies and Mendelian randomization: hands-on by Marie Verbanck and Katia Antonenko
  • Dec. 8 afternoon: Dimensionality reduction: lecture by Florian Massip and Clustering: lecture by Chloé Azencott
  • Dec. 15 afternoon: Dimensionality reduction: hands-on by Florian Massipand Katia Antonenko and Clustering: hands-on by Chloé Azencott and Katia Antonenko.