SPECTRAL VS. STATISTIC APPROACH OF STRUCTURE–ACTIVITY RELATIONSHIP. APPLICATION ON ECOTOXICITY OF ALIPHATIC AMINES

Abstract

In the context of quantitative structure–activity relationship (QSAR) studies the traditional statistical indices, i.e. simple correlation factor, standard error of estimates, as well as t-Student and Fisher's tests are challenged against the algebraic spectral description of the multi-linear models. While the new correlation approach is built on the algebraic (Euclidian) norms and on the associate correlation factor, the validation and predictive features are assured throughout so-called "ergodic" least action principle across the investigated models. The reliability of the present proposed spectral respecting statistical QSAR analysis is evaluated on modeling the aliphatic amines' ecotoxicity by means of hydrophobicity, polarizability, and total energy structural parameters resulting in better prediction of molecular hierarchy of observed bioactivity being susceptible to such general behavior.