An important aspect of biological macromolecules is how they interact with each other. Modeling the interaction between molecules is of great practical importance: understanding these interactions is a key element in the new field of computer-aided drug design. Unfortunately, detailed analysis of molecular interactions is computationally expensive.
When modeling the interaction between two molecules, we can improve computational efficiency by concentrating our initial efforts at understanding the interactions between the molecular surfaces, and evaluating the detailed atomic interactions later.
This strategy has advantages for simplifying the calculations and lowering the hardware requirements for displaying the results. HARMONY computes a simplified representation of protein surfaces using expansions of spherical harmonic functions. These representations are useful because it gives the user control of the resolution of the surface geometry and surface properties, and is especially important when developing efficient algorithms for computing and analyzing protein-protein and protein-ligand interactions. For example, the protein-protein docking program SurfDock uses low-resolution models for initially placing the molecules and high-resolution models for refinement of the complex. Details of the mathematical methods used by HARMONY can be found in Duncan, B. S. and Olson, A. J. Approximation and characterization of molecular surfaces. Biopolymers 33, 219-229, 1993.
This document - still in development - gives an introduction to HARMONY. The images were rendered using AVS, a data-flow visualization system, and AVStool, a programmable interface to AVS.