Multiple Formula Approach for Structure-Cytotoxicity/Antiviral Activity Relationship Studies of Nucleoside Analogs

Predictive Toxicology of Chemicals: Experiences and Impact of AI Tools

Volume

Predictive Toxicology of Chemicals: Experiences and Impact of AI Tools

Contents

Quantitative structure-activity relationships (QSAR) were developed for a series of purine nucleoside analogs with antiviral activity. The correlations of chemical structure of these purine nucleoside analogs to their toxicity/activity were investigated using molecular similarity analysis. Structure-activity relationship studies and molecular similarity analyses were performed using the molecular descriptors, number of atoms and bonds of a molecule (NAB), maximum common substructure (MaCS), and molecular similarity index (MSI). The antiviral activity measurement used in this study was the 50% effective dose (ED50) in (M. The cytotoxicity measurement used in this study was the 50% cytotoxic dose (CD50) in (M. The biological activities and MSI were utilized to generate a series of correlation equations. The multiple formula approach (MuFA) used the top regression correlation equations, based on several reference compounds, to generate the average estimated CD50 and ED50 values for a set of testing compounds. The MuFA integrated the effects of structural similarities and dissimilarities in estimating the cytotoxicity and antiviral activity of testing compounds.