Skip to main content

Auguste Genovesio - équipe en émergence

Computational biology

The research project of our team is the study of cellular morphology and dynamics at large scale. We are interested in characterizing the morphological heterogeneity of cellular responses to perturbations. We thus work to identify mechanical or molecular factors of cell morphology, organization and activity in different contexts. In this perspective, we contribute to generate large data sets of images or gene expression. On the other hand, we interpret these large data sets to produce and validate predictive models. As the scale of the data produced this way constrains us to full automation and quantitative approaches, we develop algorithms and tools for the analysis of large image data. The members of our team bring together a wide range of expertise such as computer science, applied mathematics, biophysics and genomic analysis. We apply our approaches to scientific questions we raise such as understanding the action of small compounds with the support of our collaborators from the Curie Institute next door and the pharmaceutical industry. We also develop approaches dedicated to fundamental biology research through a strong interaction with our colleagues at IBENS, Collège de France and ESPCI in various subfields such as functional genomics, developmental biology and neuroscience.

The research activity of the team can be roughly divided into three main themes, among which there are many bridges: the first theme concerns the study of the set of cellular morphologies obtained when cells in culture are subjected to a large number of parallel perturbations, the second theme concerns the study of the spatial relationships between cell phenotypes and their dynamics within tissues. Finally, the third theme concerns the development of novel approaches to analyse new types of data resulting from high-throughput sequencing.

Cellular morphology on a large scale. The data sources corresponding to this theme are large sets of images (typically several hundreds of thousands). These are the result of acquisitions by automated microscopy of a set of parallel perturbations (typically several tens of thousands) on one or more cell types in culture. Obtaining images containing tens of millions of cells presses us to explore ranges of cell phenotypes and the relationships between them as well as their possible molecular sources. We are interested in the phenotypic profiles of each cell to characterize groups of perturbations or to study the heterogeneity of those perturbations. We are also interested in deep learning approaches for complex images such as neuronal cultures for which the detection of individual cells is impossible. We propose and develop several approaches to identify the mechanism of action of a therapeutic molecule.

Spatial structure and tissue dynamics. We are also interested in the analysis of large image data corresponding to tissue samples containing several hundred thousand developing cells. We develop detection algorithms and original statistical approaches based, among other things, on the generation of synthesized images in order to identify zones of the tissue where spatial location of cells is not random but rather the result of a Mechanical or chemical process that we then try to understand. We also study the dynamic aspect of the activity of organ-forming neuronal cell populations, in particular for the in vivo monitoring of Kenyon cell activity for the study of long-term memory formation

Analyses of new types of high-throughput sequencing data. New sequencing technologies and new types of NGS data appear regularly. We are working on the development of analytical methods associated with these new sequencing technologies such as single molecule sequencing PacBio or nanocanals BioNano. We also develop appropriate frameworks for the analysis of new NGS data types such as Ribosome profiling, ATAC seq, Clip seq or single scRNA seq and scChip seq scans. The objective of this activity is to obtain molecular data related to the first two themes. We hope to establish a general relationship between a graph of molecular regulation and a transition graph of the morphological states of the cell.

From cell images to information
From cell images to information