SIMILARITY SEARCHING IN FILES OF 3-DIMENSIONAL CHEMICAL STRUCTURES - EVALUATION OF THE EVA DESCRIPTOR AND COMBINATION OF RANKINGS USING DATA FUSION

EVA is a new molecular descriptor that provides a concise summary of t he fundamental frequency components of a molecule's infrared range vib rational spectrum in a vector format. Target structures from the Starl ist database are used to demonstrate the effectiveness of the descript or for similarity searching and its difference from a conventional sim ilarity measure based on the matching of two-dimensional (2D) fingerpr ints. The use of data fusion on the rankings resulting from the EVA-ba sed and the 2D-based similarity measures results in a combined ranking that can be more effective in simulated property prediction experimen ts than either of the individual rankings.