Thèse soutenue

Exploration des interactions peptide-protéine
FR  |  
EN
Accès à la thèse
Auteur / Autrice : Pierre Thevenet
Direction : Pierre Tuffery
Type : Thèse de doctorat
Discipline(s) : Bioinformatique, Analyse de génome et modélisation
Date : Soutenance en 2014
Etablissement(s) : Paris 7

Mots clés

FR

Mots clés contrôlés

Résumé

FR  |  
EN

Thanks to high-throughput sequencing, many genomic sequences are identified as capable of expressing peptides. The naturel products extractions techniques expose a large number of peptides expressed, such as venom peptide. Ln addition, peptides whose sequence is derived from protein would better characterize and monitor protein-protein interactions. However, experitnental methods do not solve the structures of peptides on a large scale, as shown by the small amount of peptide structures present in the PDB. Therefore, in silico modeling is crucial to fil this gap. Homology modeling, requiring rigid reference structures, is not suitable fo peptides for which the known structures are a few. This is why de novo modeling methods were developed. My thesis focused on the adaptation of PEP-FOLD, a method for de novo prediction of peptide structures, for large seule modeling. A first point was th consideration of the disulfide bridges, commonly found in natural peptides, but also used as the stabilizing elements of the structures of peptides in the context of their engineering. A second point was to focus on optimizing PEP-FOLD. The development of a new algorithm has accelerated computing time by a factor of ten, opening the possibility to structure prediction of millions of sequences, The applications stemming from this improvement are interactions between peptides proteins, and modeling novo protein fragments of smaller sizes, difficult to model by homology, such as N and C extemities, and the linker inter-domains.