Structure-activity studies of musk odorants using pattern recognition: bicyclo- and tricyclo-benzenoids

Abstract
The relationships between molecular structure and the musk odor quality were investigated using pattern recognition techniques. A set of 148 bicyclo- and tricyclo-benzenoid compounds (67 musks, 81 nonmusks) have been studied. A discriminant function developed from a set of 14 molecular structure descriptors correctly assigned every compound of the training set into its respective category. This discriminant has also been validated using an external prediction set of 16 compounds.