In this paper we report results on the problem of docking two large proteins by means of a two-phase monotonic basin hopping method. Given an appropriate force field which is used to measure the interaction energy between two biomolecules which are considered as rigid bodies, we used a randomized global optimization methods based upon the repeated use of local searches. These local searches include two phases, the first of which is aimed towards obtaining a close match between the two interacting bodies by means of the inclusion of a penalty term in the energy, the reduction of the van der Waals radii of the atoms and the elimination of the electrostatic contributions. A single relatively small test example was chosen and used to tune the parameters of our method; the set of constants found this way was then consistently used to dock several other complexes obtained from the Brookhaven Protein Database. The results are very encouraging as in most cases the correct docking was eventually found and those cases in which a failure of the method was observed were later on discovered to possess characteristics which are incompatible with the hypotheses on which the experiments were made. It is clear to the authors that rigid docking is only a rough approximation of what is really needed, i.e. flexible docking. However, in the case of protein-protein docking it has been observed by various authors that when two large molecules interact, the change in atom positions within the same molecule just occurs in the interface between the two. So it can be the case that rigid docking might prove a very valuable tool as a first phase in which two large molecules are put as close as possible. Flexible refinements can then be performed using the results of rigid docking as starting configurations. Some initial experiments within this context display quite promising results.
Citation
Technical report, DSI, 2003