Molecular Insights into Active Phytochemicals for Diabetes and Inflammation Treatment: A Web App-Based Machine Learning and Network Pharmacology Approach

Ajay Manaithiya,Ratul Bhowmik,Ahmed A. Elhenawy,Sameer Sharma,Susha Dinesh,Seppo Parkkila,Ashok Aspatwar

DOI: https://doi.org/10.1101/2024.05.05.592582

2024-05-05

Abstract:Traditional medicinal plants such as show promising activities against diabetes and inflammation, yet their underlying molecular mechanisms remain elusive. Nevertheless, further elucidation of the molecular mechanism by which this medicinal plant exerts its anti-diabetic and anti-inflammatory effects and the relationship between its active constituents’ structure and activity is needed. Using a polypharmacology approach, this study identified metabolic pathways targeted by active phytochemicals of the medicinal plant . Extensive network pharmacology and systems biology approaches, including protein-protein interaction network analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, and gene functional enrichment studies, revealed that AKT1 is the key gene involved in the molecular mechanism of active phytochemicals involved in their antidiabetic and anti-inflammatory effects. A robust machine-learning-guided bioactivity prediction model-based web application ( ) against AKT1 was developed by implementing PubChem and substructure fingerprint molecular signatures. Further validation was performed by conducting ROC and applicability domain analysis, with subsequent molecular docking studies to understand the molecular mechanisms of the phytochemicals. The web application ( ) predicted that were the most active phytochemicals responsible for the antidiabetic and anti-inflammatory effects of the medicinal plant , which was further supported by extensive molecular docking analysis and molecular dynamics simulation studies. These findings indicate a correlation between the structure of these molecules and their bioactivity, with some phytochemicals performing comparably or better than known FDA drugs. The results show that natural products could be very useful for medicine. They encourage more lab and animal studies and lay a strong base for future research on how these natural products can lead to new drugs.

Bioinformatics