Abstract
The aim of this review is description of the logic and evolution of optimal descriptors OCWLGI calculated with the molecular graph and the demonstration of their ability as tools for the modeling of biological and physicochemical parameters of chemical compounds. The ability of optimal descriptors calculated with hydrogen suppressed graph (HSG), hydrogen filled graph (HFG) and graph of atomic orbitals (GAO) is demonstrated as a collection of quantitative structure-property relationships (QSPR) and quantitative structure-activity relationships (QSAR) for properties and endpoints available from the literature. The Monte Carlo method optimization of the correlation weights of local and global invariants (OCWLGI) of molecular graphs is used as the principle for building up descriptors which are discussed in this article. The statistical quality of the QSPR and QSAR models for physicochemical and biological properties which were obtained with the optimal descriptors are reasonably high.
Keywords: Global invariant, local invariant, Monte Carlo method, optimization of correlation weights of local and global invariants of graph (OCWLGI), quantitative structure-property relationships, QSPR, quantitative structure-activity relationships, QSAR.
14