Efficient algorithms for modeling macromolecular interactions

The biochemical activity of many proteins depends on interactions with one or more molecular partners. Many of these interactions occur in weak, transient complexes, making structural analysis by experimental methods difficult. For this reason, computational prediction of complex structures is an invaluable tool for its potential utility in determining these molecular interactions.;A protein-protein docking program, Piper, has been developed. Piper uses Fast Fourier Transforms (FFT) to efficiently sample the translational space using cross-correlations of simplified receptor and ligand energy functions. Together with rotation, a discretized six-dimensional conformational space of the receptor and ligand proteins can be quickly sampled. The principal novelty of Piper is its simple representation of a complex energy function, which provides improved results compared to traditional methods.;One of the terms in the energy function is a structure-based pairwise protein-protein interaction potential. This potential has been created within the framework of the Boltzmann assumption, which says that frequently observed structural features correspond to low-energy states. A critical element of such potential functions is the reference state, representing background energetic interactions. The novel use of atom-type-independent docking decoys as the reference state greatly improves the protein docking results.;Proteins bind not only to other proteins, but also---among other things---to small molecules in the cell. Experimental studies suggest that the binding sites generally contain smaller regions that provide major contributions to the binding free energy---often termed "hot-spot" regions. An FFT-based algorithm, FTMAP, has been developed to determine these hot-spots computationally rather than experimentally. FTMAP has been applied to a number of different protein systems for which there is experimental data. According to these results, FTMAP quickly and reliably identifies the hot-spot regions of a protein binding site.