Thèse de doctorat
Résumé : Protein structure is not fixed in time, and conformational transitions are the keyto many biological interactions such as enzymatic reactions or signal transduction.Protein dynamics and secondary structure propensities describe these conformationaltransitions, defining how protein structure is likely to evolve in time. Unfortunatelythis kind of information is extremely hard to obtain and the required experimentsare expensive and time consuming.A backbone dynamics and secondary structure propensity predictor that worksfrom sequence only, called DynaMine, has recently been developed. DynaMine addsa new dimension to protein sequences that can, for instance, be exploited to identifyprotein regions involved in specific biological tasks and in protein classification. Thisthesis shows how these predictions can be used to infer emergent properties of pro-teins and how they can highlight hidden evolutionary relationships between remotehomologs.The thesis is divided in 4 parts, corresponding to four different topics in which dy-namics and secondary structure propensities are reported to play a crucial role: thefirst part describes a new pairwise algorithm that uses secondary structure propen-sities and dynamics to align remote homologous proteins. The following three partsdeal with dynamics-related protein emergent behaviours: protein disorder, beta-aggregation and DNA-binding capability in archaea. The results show how backbonedynamics and secondary structure propensities can help improving the prediction ofall the aforementioned subjects. With regard to the identification of DNA-bindingproteins and the prediction of beta-aggregation, experimental validation is also pro-vided and discussed.