Proteins are essential parts in living organisms. They participate in most of the cellular processes such as gene expression, signal transmission, catalysis of chemical reactions, … Due to their large range of possible functions, the study of proteins interests other fields in addition to biology. Proteins are pharmaceutical targets and drugs, their catalytic properties are widely used in biotechnology, and they are used as components of nano-devises in the rising field of bionanotechnology. Although the properties of natural proteins can be directly exploited, new, designed proteins, with novel functions or improved activities, are of major interest in all these application areas. Protein design may involve the remodeling of a known protein scaffold in order to modify the protein function/activity, or, in the most general case, the complete (de novo) design of new protein structures to fulfill a particular function. The problem is extremely challenging since the number of possible combinations of amino acids to be tested is astronomically large. Experimentally testing all the possible sequences is practically impossible. Therefore, computational protein design methods have been developed for over a decade. In addition to the intrinsic combinatorial complexity of the protein design problem, computational methods have to face the natural flexibility of proteins (i.e. proteins are flexible molecules that fluctuate between nearly isoenergetic states). Indeed, the protein design problem is even more challenging if dynamical aspects (e.g. allosteric shifts, loop motions, ...) are considered in addition to static aspects (e.g. positional arrangement of catalytic residues for enzyme activity). Due to all these difficulties, and despite great advances in recent years, computational protein design remains a largely open problem. In particular, improvements in models and algorithms are essential to better explore the protein sequence combinatorial space while taking into account protein flexibility. Besides, accurate and computationally efficient energy functions, able to better account for interactions with solvent and entropy change, are necessary. The goal of this project is to yield advances in a general methodology for protein design, and to develop suitable computational design tools that will lead the development of new proteins for applications in biotechnology, biomolecular nanotechnology, molecular medicine and synthetic biology. The methodological breakthrough expected from this interdisciplinary project builds on the combination of cutting-edge methods in computational biology with efficient algorithms originating from robotics. Among all the possible applications of the methods developed in this project, special attention will be given to enzyme design for applications in biotechnology such as the production of high-valued molecules, the development of eco-friendly bioprocesses and the valorization of renewable carbon resources. Such applications are of high interest to the pre-industrial demonstrator Toulouse White Biotech (TWB), supporter of our project, and to the Competitive Cluster AgriMip. The achievement of the project relies on the complementary expertise of four partners: LAAS-CNRS for robotics and computer science, BIOS-Polytechnique and LISBP-INSA for computational biology & protein engineering, and Kineo CAM, a company specialized in software development for computer-aided design and manufacturing.