Predicting peroxisome proliferator-activated receptor gamma potency of small molecules: a synergistic consensus model and deep learning binding affinity approach powered by Enalos Cloud Platform.

Abstract

Peroxisome proliferator-activated receptor gamma (PPARγ) antagonists play a critical role in regulating glucose and lipid metabolism, making them promising candidates for antidiabetic therapies. To support the ongoing search of such compounds, this study introduces two advanced in silico models for predicting the binding affinity and biological activity of small molecules targeting PPARγ. A neural network was developed to classify compounds as strong or weak binders based on molecular docking scores. Additionally, a consensus model combining Random Forest, Support Vector Machine, and k-Nearest Neighbours algorithms was implemented to predict the antagonistic activity of small molecules. Both models were rigorously validated according to the Organisation for Economic Co-operation and Development (OECD) guidelines, to ensure generalisability and sufficient efficiency in detecting the minority class (active antagonists). Mechanistic insights into how key molecular descriptors influence PPARγ activity were discussed in a posteriori interpretation. A case study involving 34 prioritised per- and polyfluoroalkyl substances (PFAS) were screened with the developed workflows to demonstrate their practical application. The models, integrated into user-friendly web applications via the Enalos Cloud Platform, enable accessible and efficient virtual screening, supporting the discovery of PPARγ modulators.