Please login first
Multivariate classification of a series of organic compounds of pharmaceutical interest using MODESLAB methodology
* 1 , 2 , 3 , 1
1  Department of Pharmacy. Institute of Pharmacy and Foods. University of the Havana
2  artment of Pharmacy. Institute of Pharmacy and Foods. University of the Havana
3  Company Laboratories AICA


Due to the high cost of the development of new active pharmaceutical ingredients and excipients for the pharmaceutical industry, molecular modeling methods have been included in this process more and more frequently in recent years. In this work the calculation of the spectral moments of the matrix of adjacency between edges of the molecular graph with suppressed hydrogens was made, weighted in the main diagonal with different parameters that characterize both the bonds and the atoms in the molecules of compounds of pharmaceutical interest, using the MODESLAB methodology. 91 descriptors related to solubility were calculated, which were used in a training series divided into five groups, according to the priority rules of the IUPAC. With the aim of identifying the descriptors that best discriminate between the compounds of each group and defining the set of functions of these descriptors able to distinguish with the greatest possible precision the members of one or the other group, a discriminant analysis was developed using the stepwise inclusion method using the statistical software IBM SPSS version 22. Four functions were generated that constitute combinations linear of 16 molecular descriptors, which encode both steric and electronic information of the molecules of each group. The functions obtained have a very low minimum Wilks Lambda (0.067) and a high canonical correlation (0.89), which demonstrates their discriminant power and allows the use of the descriptors included in them in future studies of structure-property or structure-relationship. activity.

Keywords: structure-property structure-relationship. activity. descriptors