À la Une

Soutenance de thèse Thomas Charlon


M. Thomas Charlon soutiendra en anglais, en vue de l'obtention du grade de docteur ès sciences mention informatique de la Faculté des sciences de l'Université de Genève, sa thèse intitulée:

Genetic Clustering for the Discovery of a New Classification of Systematic Autoimmune Diseases

Date: Lundi 2 septembre 2019 à 15h00

Lieu: CUI / Battelle bâtiment A, Auditoire rez-de-chaussée



  • Prof. Sviatoslav Voloshynovskyi (Directeur, Université de Genève)
  • Prof. Zoltan Kutalik (Génétique Médicale, Lausanne)
  • Dr. Frédérique Lisacek (SIB, Université de Genève)
  • Dr. Markus Müller (SIB, Lausanne)
  • Dr. Rostyslav Kuzyakiv (Informatique, Zürich)


Systemic autoimmune diseases have serious clinical consequences, affect 1% of the global population and have limited treatment options. Although risk factors have been already identified, clinicians expect that the current classification of the diseases could benefit from a molecular-based reclassification, to better explain their partly shared genetic contributions. In that objective, more than 1,000 genotypes of systemic autoimmune diseases patients were measured to perform clustering of patients altogether, characterize the clusters and identify genetic signatures.
In this thesis, two complementary approaches were used to reclassify the patients: genome-wide, in which all the hundreds of thousands of markers are used, and candidate-based, in which only the markers known to be associated with the diseases, as reported by previous association studies, are used. The state-of-the-art is extended in several important ways. First, a novel genome-wide selection and summarization method is described and evaluated, and features are then clustered using a density-based clustering workflow. Then, the candidate-based approach is performed using Gaussian mixture models and reveals novel insights about subtypes and symptoms shared among diseases. Finally, to increase the quality and robustness of the candidate-based clustering, sparse coding feature transformation methods are evaluated and compared.
The newly developed methods enabled to find disease relevant clusters using genome-wide markers and enabled a precise description of expected and novel profiles using disease associated markers. This thesis thus describes novel genetic clustering methods that are useful for the molecular reclassification of diseases and reveals novel insights on the partly shared genetic contributions of systemic autoimmune diseases.