Novel atomic-level-based AI topological descriptors: application to QSPR/QSAR modeling.

Abstract

Novel atomic level AI topological indexes based on the adjacency matrix and distance matrix of a graph is used to code the structural environment of each atomic type in a molecule. These AI indexes, along with Xu index, are successfully extended to compounds with heteroatoms in terms of novel vertex degree v(m), which is derived from the valence connectivity delta(v) of Kier-Hall to resolve the differentiation of heteroatoms in molecular graphs. The multiple linear regression (MLR) is used to develop the structure-property/activity models based on the modified Xu and AI indices. The efficiency of these indices is verified by high quality QSPR/QSAR models obtained for several representative physical properties and biological activities of several data sets of alcohols with a wide range of non-hydrogen atoms. The results indicate that the physical properties studied are dominated by molecular size, but other atomic types or groups have small influences dependent on the studied properties. Among all atomic types, -OH groups seem to be most important due to hydrogen-bonding interactions. On the contrary, -OH groups play a dominant role in biological activities studied, although molecular size is also an important factor. These results indicate that both Xu and AI indices are useful model parameters for QSPR/QSAR analysis of complex compounds.