Recent uses of topological indices in the development of in silico ADMET models

Abstract

Topological indices are employed in an ever-widening family of quantitative models to describe the structural attributes of compounds as these relate to experimental endpoints in a host of physicochemical and biological processes. This is especially true where attention to ADMET (absorption, distribution, metabolism, excretion and toxicity) properties is a priority, using various training or learning algorithms to construct quantitative structure-activity or -property relationship ADMET models. This review discusses the in silico ADMET approaches used over the past two years, where the majority of descriptors are topological, including comparisons between models and their important descriptors where applicable. ADMET models for aqueous solubility involving several large datasets are reviewed, as are a number of models for human intestinal absorption. Also included is the use of topological indices extended to modeling metabolic stability for the cytochrome P450 cassette of enzymes in an interesting study with compounds measured in a uniform bioassay using human liver S9 homogenate. Finally, in the area of genotoxicity, two recent ADMET models are discussed, one for chromosomal aberrations and another using a large compound dataset for Ames mutagenicity. The latter study involved several thousand compounds, with comparisons of validation results for a number of well-known predictors of mutagenicity.