Chapter 15 A quantitative structure-activity relationship of 1,4-dihydropyridine calcium channel blockers with electronic descriptors produced by quantum chemical topology

Abstract

Abstract Over the years, quantum topological molecular similarity (QTMS) has been developed and used successfully in the framework of quantitative structure-activity relationships (QSARs). In time we accumulated considerable evidence that (geometry) optimised ab initio bond lengths supplemented with bond critical point properties function as reliable descriptors capturing electronic effects. In this article, this assertion is tested in the context of the well-known and most challenging medicinal QSAR of 1,4-dihydropyridine calcium channel blockers. The complexity of this QSAR is due to the varying influence of lipophilic, steric and electronic descriptors depending on the position of the substituent (ortho, meta and para). Four types of chemometric analysis were applied to QTMS descriptors generated at AM1, HF/3-21G* or HF/6-31G* level of theory: a standard partial least square (PLS) analysis of the whole data set (as well as for the ortho, meta and para subgroups alone), a genetic algorithm (GA) analysis of raw ‘variables’ followed by a PLS regression (GA-PLS), an artificial neural network (ANN) operating on principal components (PCs) of GA-selected variables (PCA-ANN) and a self-organised map (SOM) or Kohonen neural net. Quantum topological descriptors are shown to be sound substitutes of well-established classical empirical parameters such as the Hammett constant, even in convoluted QSARs where nonelectronic parameters feature strongly. It is valuable that modern quantum chemistry can now routinely deliver new well-defined descriptors to be used in complex biological QSARs.