Protein-Protein docking is a recent practice in biological research which involves using 3D models of proteins to predict the structure of complexes formed by these proteins. Studying protein-protein interactions and how proteins form molecular complexes allows researchers to better understand their function in the cell. Currently, the most common methods used for docking are fully computational approaches, followed by the use of molecular visualization tools to evaluate results. However, these approaches are time consuming and provide a large number of potential solutions. Our basic hypothesis is that a virtual reality (VR) framework for molecular docking can combine the benefits of multimodal rendering, of the biologist's expertise in the field of docking, and of automated docking algorithms. We think this approach will increase efficiency in reaching the solution of a docking problem. However designing immersive and multimodal virtual environments (VE) based on VR technology calls for clear and early identification of user needs. To this end, we have analyzed the task of protein-protein docking as it is carried out today, in order to identify benefits and shortcomings of existing tools, and support the design of new interactive paradigms. Using these results, we have defined a new approach and designed a multimodal application for molecular docking in a virtual reality context.