Habilitation thesis of Maude Pupin

Computational biology for nonribosomal peptides and their synthetases

In the thesis of my "Habilitation à diriger des recherches", I'm presenting the pioneer work on computational biology for nonribosomal peptides (NRPs). Those researches began in 2006 in Lille and lead to the unique plate-form dedicated to computational biology analysis of NRPs called Norine, of which I am a founder member. Nonribosomal peptides are small molecules produced by micro-organisms, bacteria and fungi, to colonize their environment. Those peptides have the advantage of presenting a large range of structures. They can be linear, but also can contain cycles and/or branches, and consist of more than 500 different building blocks. This variety comes from their synthesis done by huge enzymatic complexes, called nonribosomal peptide synthetases (NRPSs). They select amino acids and other compounds, called monomers, and then link them by peptide and other bonds. So, nonribosomal peptides cover a large range of activities such as antibiotic, anti-tumor or immuno-suppressor. Some, as penicillin, are commonly used as drugs. In the first part, I present the synthetases by associating the peptides' characteristics to the enzymatic functions required to accomplish them. Then, I describe the main steps needed to design a tool that analyse the NRPS protein sequences by specifying the characteristics of the existing tools. Then, I present my own contribution to the investigation of the production of NRPs from genomic or protein sequences by participating to the design of bioinformatics protocols and to genome annotation. In the second part, I start by specifying the contributions of the Norine resource to the knowledge of the nonribosomal peptides diversity, supplemented by a study of the chemistry of those molecules. Next, I present the few databases and tools related to those peptides that are developed elsewhere. Then, I describe my own contribution to the Norine resource and suggest modernisation of the process to collect data and expansion of the query functionalities for structure search. I finish by suggesting new prospects: cheminformatics dedicated to nonribosomal peptides with the goal to predict one or several synthetases able to produce a peptide having a given activity.

defended on 03/12/2013