Similarity Searching in Files of Three-Dimensional Structures: Evaluation of Similarity Coefficients and Standardisation Methods for Field-Based Similarity Searching

There is increasing interest in the development of field-based, similarity-searching systems, in which a user's target structure is matched against the structures in a database to find those that are most similar to it, on the basis of some quantitative similarity measure. This paper describes an extended comparison of two of the major components of such a measure, the similarity coefficient and the standardisation method, when used with one common form of field-based data, viz. the steric and electrostatic fields used in a CoMFA analysis. Our results show that there is little difference between the performances of the various coefficients and standardisation methods when tested in simulated property-prediction experiments. Accordingly, any of the coefficients and standardisation methods are appropriate for use in a field-based similarity searching system.