Application of Linear Discriminant Analysis in the Search of Antiepileptogenic Drugs

CAROLINA L. BELLERA; ALAN TALEVI; LUIS E. BRUNO-BLANCH

ACTA FARM. BONAERENSE

Colegio de Farmacéuticos de la Provincia de Buenos Aires

Año: 2006 vol. 26 p. 244 - 252

0326-2383

TradiTional pharmacologic treatment of epilepsy is aimed to the control of convulsions that characterize epilepsy crisis. In contrast, the latest research in the epilepsy field points towards the understanding of the mechanism involved in epileptogenesis, in order to achieve the discovery of drugs that inhibit the disease progression, minimizing the probability of developing epilepsy in those patients considered at risk. In the present work we have applied linear discriminant analysis to generate a discriminant function based on constitutional and topological descriptors and capable of differentiating drugs with and without antiepileptogenic activity. The discriminating function that was chosen on the basis of statistical parameters (Snedecor?s F, Wilk?s U and global percentage of good classifications) was validated internally and externally. Afterwards, it was applied in the selection of potentially antiepileptogenic compounds from a database of 250,000 chemical structures. Lipinski?s rule of five and the optimal value of octanol-water partition coefficient for drugs acting at the central nervous system were applied as secondary filters, selecting 677 compounds with antiepileptogenic potential.