Prev. Article Next Article Table of Contents

Research Article

Generation and Display of Activity-Weighted Chemical Hyperstructures

Nathan Brown, Peter Willett,* and David J. Wilton

Krebs Institute for Biomolecular Research, Department of Information Studies, University of Sheffield, Western Bank, Sheffield, S10 2TN, U.K.

Richard A. Lewis

Eli Lilly & Company, Erl Wood Manor, Windlesham, Surrey, GU20 6PH, U.K.

J. Chem. Inf. Comput. Sci., 2003, 43 (1), pp 288–297

DOI: 10.1021/ci0103875

Publication Date (Web): January 27, 2003

Corresponding author phone: +44-114-2222633; e-mail: p.willett@ sheffield.ac.uk.

Abstract

A chemical hyperstructure is a single graph representation of a set of molecules that minimizes the degree of structural redundancy in the data set. This paper describes the use of a genetic algorithm to generate an activity-weighted chemical hyperstructure (AWCH) by sequentially mapping each molecule in the data set to the hyperstructure and then assigning activity and inactivity frequency weights to the nodes and edges of the hyperstructure. Experiments with several data sets demonstrate the level of activity clustering in an AWCH.

View: Full Text HTML | Hi-Res PDF