Computational Approach to Select Lead-Like Compounds From an Opioid Class of Novel Psychoactive Substances

The continuous emergence of novel psychoactive substances (NPS) poses significant challenges to drug discovery, regulation, and public health. In this study, a computational chemometric approach was applied to evaluate 134 opioids belonging to the NPS class using ACD/Percepta software. Parameters that were predicted are physicochemical characteristics, ADME (Absorption, Distribution, Metabolism, and Excretion) profiles, and toxicity endpoints. Data pre-processing involved standardization, outlier detection via Grubbs' test, and distribution assessment using the Kolmogorov-Smirnov test. Multivariate statistical analyses (MANOVA, discriminant analysis, and principal component analysis) revealed significant correlations between Lipinski and lead-likeness violations and parameters such as log P, molecular weight, solubility, protein binding, blood-brain barrier penetration, metabolic stability, P-glycoprotein interaction, and cytochrome P450 inhibition. Discriminant analysis achieved up to 100% classification accuracy for certain parameter combinations, suggesting that log P and molecular weight alone can reliably predict broader ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) profiles. These results demonstrate the utility of integrating in silico prediction platforms with chemometric methods to prioritize candidate compounds for further experimental evaluation, thereby supporting efficient lead selection within the opioid NPS class.