A computational model to predict clathration of molecules with cephradine
Kemperman GJ, de Gelder R, Wehrens R, Dommerholt FJ, Klunder AJH, Buydens LMC, Zwanenburg B
JOURNAL OF THE CHEMICAL SOCIETY-PERKIN TRANSACTIONS 2 (6): 981-987 2001

Abstract:
The prediction of inclusion of molecules in the hosting framework of the cephalosporin antibiotic cephradine has been investigated. For this purpose linear discriminant analysis has been applied on molecular similarity data. The way the molecular similarity is calculated appeared to be extremely important. The charge distribution similarity of two molecules calculated at optimal shape overlap, is most appropriate to derive a computational model. The predictive power of the model increases when the molecular similarity of a molecule with respect to only a limited number of guiding compounds is used. The ultimate outcome is that complexing behaviour of a molecule can be predicted using a simple equation containing the similarity indices of that molecule with respect to three guiding compounds only. The model eventually obtained can predict the complexing behaviour of independent test sets of molecules with an average score of 86%.