Screening of Medicinal and Edible Homology Substances for Diabetic Kidney Disease Based on GraphBAN
Yaxian Ning, Xiaochun Zhou, Lili Zhang, Gouqin Wang, Jianqin WangBackground:
Diabetic kidney disease (DKD) is a critical microvascular complication of diabetes mellitus, and the current pharmacotherapies are limited by side effects. Medical and edible homology (MEH) Agents: The DKD and healthy control multi- -target active and low-toxic drugs (MEH) agents were identified based on transcriptomic data of DKD patients and healthy controls retrieved in the GEO database.
Methods:
The Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) and other databases were searched to obtain active components of 106 MEH substances. The GraphBAN model was used to predict the compound-target interaction and filter candidate components and genes based on the five rules proposed by Lipinski. A multi-level MEH-Ingredient-Gene-Pathway network was built to discover important active ingredients and hub genes. The protein-protein interaction (PPI) network analysis, immune cell infiltration analysis, molecular docking, and molecular dynamics simulations were further used to validate core interactions.
Results:
A total of 436 DEGs were identified. Screening via the GraphBAN model generated 6 candidate active ingredients and 11 candidate genes, from which 5 key ingredients (aurantio-obtusin, obtusin, licoisoflavanone, triptolide, triptolidenol) and 3 core genes (CERS6, CETP, FYN) were determined. PPI network analysis suggested that these core genes synergistically regulate lipid metabolism and immune processes. Further evaluation revealed concurrent immune activation and suppression in DKD, with key genes negatively correlated with immune cell infiltration levels. Finally, molecular docking and dynamics simulations verified stable binding affinities of the corresponding complexes.
Discussion:
The present study investigated the therapeutic potential of multi-target, low-toxicity MEH substances against DKD. Using transcriptomic profiling, compound-- target prediction, and regulatory network construction, we identified 5 key MEH ingredients and 3 core genes. These molecules may stabilize DKD progression via modulating lipid metabolism and immune-inflammatory pathways, providing a basis for the development of multi-target natural products and supporting the value of artificial intelligence in MEH-related research.
Conclusion:
In the current study, the authors examined the anti-DKD effects of MEH agents and identified 5 main ingredients and 3 central genes, which could prevent DKD due to the regulation of lipid metabolism. Some of the limitations, such as a rather limited sample size, must be taken up in future studies.