EFFICIENT DOCKING OF MACROMOLECULES BY EXHAUSTIVE ENUMERATION OF CONFIGURATIONS Lynn F. Ten Eyck^1,2, Igor Tsigelny¹, Victoria Roberts³, and Jeffrey Mandell^{2, 1} Department of Chemistry and Biochemistry, University of California at San Diego, La Jolla, California, ² San Diego Supercomputer Center, San Diego, California, ³ Department of Molecular Biology, The Scripps Research Institute, La Jolla, California

New approaches to macromolecular docking are being developed in several laboratories which use convolution and correlation methods to evaluate the interaction energy between two molecules over all space for any given orientation. These methods use Fast Fourier Transforms to compute the correlation functions, so they scale well as the size of the problem increases. This paper describes the development and use of a new computer program, DOT, which computes the solvated electrostatic energy and the van der Waals interaction energy for two molecules of arbitrary size, as a function of position and orientation. The computation is efficient, enumerating billions of configurations in minutes on a parallel supercomputer, or in a few hours on a group of networked workstations. Exhaustive enumertion shows all possible binding locations, within the limitation of the grid chosen. A large portion of the statistical mechanical partition function can be calculated from this data, which gives a measure of the significance of the best results. The method is well suited to screening of possibilities, with the best values being fed into a more detailed calculation for further refinement.

The use of exhaustive enumeration on a grid instead of statistical sampling in a continuum means that the method has different properties from standard energy minimization methods, and suggests different uses. DOT has been used in systems which show electrostatic steering, and systems which show tight docking. Limitations, optimal uses, and possible enhancements of the method are presented.