This project aims to develop nobels algorithms to model protein-protein complexes, a very important aspect in biophysics. The algorithm presented based only on geometrical arguments, is intended to be a first and fast approach to get the most probable configurations. The algorithm finds the best positions producing only a small number of solutions (over 250 solutions). The method is based on 2D FFT (fast fourier transform) and orthographic projections of the proteins. The method allows us to find solutions around 15 Ǻ of Cα root mean square deviation for proteins with low electrostatic interactions.