Conception d'un web service pour la fouille de données de génomique : application à la caractérisation de la myogenèse et de l'adipogenèse

摘要

The quality of carcasses and meats depends on the balance between muscle and adipose tissue (AT) masses that determine carcass weight and performance (muscle and fat composition), but also the sensory quality of the meat (tenderness, juiciness and flavor). Understanding how to control the ratio of muscle mass relative to AT mass represents a major challenge for beef producers. The balance between these masses depends on the number and volume of muscle and AT cells. These cellular events are taking place at the early steps of fetal period in cattle, as the total number of muscle cells is fixed at 180 days post-conception (dpc) in the fetus. The analysis of the evolution of these two proteome tissues during fetal life produced original but insufficient data. In addition, it is not always easy to extract or generate relevant biological information from genomic experiments. This is particularly true in ruminant species because they are not annotated in databases and few bioinformatic resources are dedicated to them. In this context, our objective was to design an “all in one” web service to analyze genomic data in cattle in order to improve knowledge of the mechanisms involved in fetal muscle and AT growth. Thus, we have organized our thesis in two axes. We developed a genomic data analysis tool, dedicated to ruminant species (cattle, sheep and goat) and named ProteINSIDE (www.proteinside.org). In a single query, this tool synthesizes the biological information stored in public databases or provided by functional annotations from gene ontology. It also predicts proteins that are secreted (tissue secretome) and which are involved in signaling between cells or tissues. It links proteins according to their molecular interactions to identify and visualize those that contribute to the same biological processes and those that are central to a biological process. ProteINSIDE was tested with data sets of 1000 proteins by species and has been successfully compared with DAVID, BioMyn, and AgBase (designed for information retrieval and annotation), as well as PrediSi and Phobius (that predict proteins secreted). We applied ProteINSIDE to the proteome analysis of muscle and AT. A first analysis of data on the ontogenesis of the tissue revealed links between proteins of both fetal tissues and proteins involved in autophagy processes. In a second study, we constructed and described the bovine proteomes of both tissues at 140 dpc. We identified 514 muscle protein and 752 AT proteins, including 346 commons proteins. As an example, these proteins are involved in the negative regulation of apoptosis, in autophagy processes, in the regulation of cell proliferation, and in the Wnt signaling pathway. We identified 47 and 93 potentially secreted proteins by muscle and TA, including 24 commons proteins. The integration of knowledges about the secreted proteins with those available for the “surfaceome” suggested proteins which could participate in the cross-talk between muscle and AT. Thus, we produced a web server to mine genomic data from bovine, sheep, and goat species, but also from human, rat and mice species. This type of server should be particularly useful to the scientific community. Its implementation has led to the production of new knowledge and working hypotheses for the understanding of the mechanisms which regulate fetal growth of muscle and AT.