After a double training in biology and computer science, I did my PhD at the Institute of Genetics and Development of Rennes (UMR-6290, CNRS-University of Rennes 1). The goal of my PhD was to use comparative genomics and population genetics to study the evolution and genetic diversity of the canine species. Indeed dog was domesticated more than 15,000 years ago and is a model for studying natural and artificial selection. Natural selection dominated the first period of dog evolution. I characterized canine genes under positive selection using a comparative genomic approach and the analysis of dN/dS ratios between the dog and nine other mammals, a project in collaboration with Hugues Roest-Croellius’s team of ENS Paris. Artificial selection led to the creation of ~ 300 modern breeds in the last centuries. I developed a population genetics and a bioinformatics pipeline starting with Fst between breed pairs to identify regions of allelic differentiation between dog breeds. This work, carried out in the context of the European consortium of canine genetics “LUPA”, made it possible to determine all regions of the genome of strong differentiations between breeds of dogs. These regions are the candidate targets for artificial selection that contribute to the high phenotypic diversity established between dog breeds. Additionally, I deployed the Fst analysis pipeline developed during my thesis in a laboratory at the University of Helsinki (Finland).
After my PhD, I joined the UMR-946 to deepen my knowledge and develop my skills in statistical and epidemiological genetics. My research mainly aimed at developing new strategies to identify new genetic factors involved in the onset and progression of melanoma, a skin cancer having an increasing incidence in Western populations. To go beyond GWAS analysis, we first proposed a multi-marker analysis strategy that integrates an analysis of biological pathways and an analysis of gene interactions within these pathways. This work, carried out in conjunction with an american team of the MD Anderson Cancer Center, allowed us to identify pathways and new genes that jointly influence the occurrence of melanoma (a study based on 6,803 subjects) and the Breslow factor index, prognosis of this cancer (2,506 subjects). I then developed inter-species comparative genomic strategies (man / dog / pig) to identify new susceptibility genes in humans. This project included several steps: Genome-wide association studies in all three species; the identification of orthologous regions between the three species; the characterization of orthologous regions associated with melanoma and the identification of sets of genes belonging to biological pathways influencing this cancer. This work, carried out in collaboration with two French CNRS and INRA teams, identified several orthologous regions associated with melanoma. During this researchs, I developed tools and computer procedures that are transmitted and used in the laboratory. Additionally, I contributed to the study projects of multifactorial diseases including asthma.
In order to continue to contribute to the understanding the genetic factors involved in complex traits by integrating different approaches, I recently joined the group of statistical genetics of the Institut Pasteur. My current objective is decipher genetic mechanisms using GWASs summary statistics by merging different analytical techniques. I focus on two complementary anthropometric traits: Body Mass Index (BMI – proxy of fat content of the body) and Wiast-to-Hip Ratio (WHR – indicator of fat distribution). These two traits are highly correlated at the phenotypic and at the genetic level in women and in men and GWAS summary statistics from the GIANT consortium are publicly available. To investigate causal relationship between these two phenotypes, several recent publications used the Mendelian Randomization (MR) principle. This approach is based on the comparison of effect estimates of SNPs on two traits. In this project, we extend this approach to characterize more finely the potential causal scenario between two phenotype. This will lead to (i) gain new insight on the genetic mechanism influencing BMI and WHR and (ii) develop a more general method for causal inference on GWAS statistics.