Proteins are involved in many essential cellular processes of living organisms. Proteins form macro complexes joining themselves to other proteins to carry out these processes. Therefore, to know the 3D structures of such complexes is of biomedical interest. Proteinprotein docking algorithms aim to predict how two proteins interact with each other to form a 3D complex. Docking algorithms need to fulfill two main tasks: (1) sampling all the possible relative positions of the two proteins and (2) computing the interaction energy at each position to find the minimum energy (= the best solution). Obtaining the interaction energy is a computationally expensive task. We are developing a new algorithm based on the ATTRACT coarsegrained forcefield [1] and using a quaternion ball representation to accelerate the search of the 3D rotational space.