The work presented in this paper corresponds to eight years of research on the problem of interaction between proteins. I approached the problem of predicting protein-protein interactions in terms of the supervised learning. I am interested in learning predictive models for protein-protein interactions. I studied two different types of interactions: - The protein-protein interaction from the protein interaction network point of view; - Physical interaction between proteins by predicting which residues are actually in interaction (protein-protein docking). The paper is structured as follows: after presenting the problem of supervised learning and unsupervised in the first chapter, the second chapter concern the prediction of protein-protein interaction. The results on the protein-protein docking is presented in the third chapter and finally, perspectives are presented in the last chapter.
