A Novel System for Assessment of Macromolecular Interaction in Condensed Phases. 2) Interaction Site Inference by Molecular Shape and Electrostatic Complementarity
Atsushi Yoshimori, Carlos Adriel Del Carpio Munoz · Proceedings Genome Informatics Workshop/Genome informatics · 1999
The authors are involved in the development of the system for automatic assessment of macromolecular interaction MIAX described elsewhere [1]. In the present study we undertake the problem of reduction of the huge configuration search space for the complex in order to improve processing times of MIAX. Here we describe an algorithm to infer interaction sites for the monomers composing the complex which dictate plausible starting points for the optimization process performed by MIAX. The algorithm is based on the computation of macromolecular complementarity, geometrical as well as electrophysical. Shape complementarity is calculated by a modification of the algorithm proposed by Katchalski-Katzir et al. [2]. Electrophysical complementarity is evaluated by means of a self organizing map (SOM) [3] algorithm which renders a nonlinear projection of the high dimensional configuration of the electrostatic potentials on the surface of the monomers into a lower dimensional map of neurons. We evaluate the algorithm predicting plausible interaction sites for several complexes whose coordinates are found in the PDB, and comparing them with the inferred candidate interaction regions.