ABSTRACT
Mathematical models were developed for the estimation of human carbonic anhydrase (CA) II inhibition. A large set of 95 CA inhibitors incorporating diverse aromatic rings were used for this purpose. The numerical descriptors used were distance- and connectivity- based indices, quantum -theoretical descriptors and Balaban and Balaban type descriptors of molecular structure. After descriptor generation, multiple linear regression analysis was performed to find superior models for estimation. The obtained results indicate that: (i) models based on topological indices are superior to those based on quantum -theoretical descriptors; (ii) combinations of topological and quantum-theoretical descriptors improves the quality of regression; (iii) in both cases involvement of Balaban and Balaban type indices is beneficial. The results are described critically based on variety of statistical parameters.
