Exploring the Functional Food Potential of Zea mays Using Machine Learning–Based QSAR and Network Biology to Identify Anti-Diabetic and Anti-Inflammatory Phytochemicals

This repository contains the datasets, code, and supplementary materials associated with the study titled “Exploring the Functional Food Potential of Zea mays Using Machine Learning–Based QSAR and Network Biology to Identify Anti-Diabetic and Anti-Inflammatory Phytochemicals.” The dataset includes curated Zea mays phytochemical structures, descriptor matrices (CSV format), and machine learning scripts used for QSAR model construction and validation. Network pharmacology data files, molecular docking outputs, and figure assets are also provided. A machine learning–guided QSAR model was developed using PubChem substructure fingerprints to predict anti-diabetic and anti-inflammatory activity, validated via ROC and applicability domain analyses. Network and pathway enrichment analyses identified AKT1 as a central target mediating the bioactivity of Zea mays. The study integrates cheminformatics, molecular modelling, and systems biology to provide mechanistic insight into maize-derived compounds. Web application: https://akt1-pred.streamlit.app Keywords: Zea mays, AKT1, Network Pharmacology, QSAR, Machine Learning, Diabetes, Inflammation, Natural Products, Drug Discovery