Structure-activity relationship studies of substituted 17 alpha-acetoxyprogesterone hormones

Recently a new methodology, called electronic indices methodology (EIM), ba sed on local density of tate calculations (LDOS) using topological and semi empirical methods, was proposed to identify the biological activity of poly cyclic aromatic hydrocarbons (PAHs) In this work we apply the concepts of t he EIM approach to,classify the progestational activity of 21,17 alpha -ace toxyprogesterones (steroid hormones), (APs). The EIM approach pointed to ra few descriptors, which correctly classify the active/inactive compounds of this class (approximate to 90%). We show that these descriptors arise natu rally from principal component analysis (PCA) and neural network (NN) calcu lations. Moreover, using only the parameters from EIM, instead of a large s et of descriptors that have been used before to describe the biological act ivity of these hormones, we slightly improve and simplify,PCA and NN result s. Finally, the molecular region related to the chemical activity of these hormones naturally appears in our theoretical analyis, from the local densi ty of states of the frontier orbitals. This shows the generality of the pri nciples of EIM approach, and confirms that the combination df these distinc t methodologies can be an efficient and powerful tool in the structure-acti vity studies of many classes of compounds.