The broad goal of this project is to develop algorithms and software for modeling the binding of a protein molecule with the molecule to which it attaches (ligand). Given the 3-D structures of the receptor and the ligand, rapid and reliable computational methods to locate the binding sites and predict the geometry of the complex will be developed. The first specific goal will be to adapt existing Fast Fourier Transform algorithms developed for the translational searches to use free energy targets for improved predictive reliability. The second goal is to develop Fast Fourier Transform methods for rotational searches as well. Finally, the methods will be tested and benchmarked on a database of protein complexes. Based on the outcome, the database along with the algorithms will be will be made available via the World Wide Web. For researchers who submit monomeric data, the web server will run the parallel version of the FFT docking algorithms on the supercomputers in the Center for Computational Science at Boston University.