|
| Identification of key genes in renal fibrosis and prediction of targeted traditional Chinese medicine |
| Hits 165 Download times 52 Received:January 25, 2026 |
| View Full Text View/Add Comment Download reader |
| DOI
10.11656/j.issn.1673-9043.2026.05.08 |
| Key Words
renal fibrosis;machine learning;feature genes;immune infiltration;traditional Chinese medicine |
| Author Name | Affiliation | E-mail | | KANG Yi | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China
Beijing University of Chinese Medicine, Beijing 100029, China | | | JIN Qian | Beijing University of Chinese Medicine, Beijing 100029, China | | | WANG Xuezhe | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China
Beijing University of Chinese Medicine, Beijing 100029, China | | | JIN Xinyan | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China
Beijing University of Chinese Medicine, Beijing 100029, China | | | LI Zirong | Beijing University of Chinese Medicine, Beijing 100029, China | | | ZHOU Mengqi | Beijing Puren Hospital, Beijing 100062, China | | | LI Xiaowen | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China | | | WANG Yaoxian | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China | | | LYU Jie | Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700, China | lvjiebucm@163.com |
|
| Abstract
|
| [Objective] To identify key genes involved in renal fibrosis（RF）using bioinformatics methods and to explore herbs that regulate these genes. [Methods] The GSE76882 dataset from the GEO database was used as the training set and GSE22459 as the validation set. Differentially expressed genes（DEGs）and key modules identified through weighted gene co-expression network analysis（WGCNA）were intersected，and protein-protein interaction networks along with enrichment analyses were conducted. Seven machine learning models were constructed to identify feature genes，and a nomogram was generated. The models were evaluated using calibration and decision curves，leading to the identification of RF-related genes. Immune infiltration analysis was performed to explore correlations between feature genes and immune cells. Based on these genes，potential traditional Chinese medicine （TCM）herbs and active compounds were predicted using Coremine and TCMSP databases，and molecular docking was performed between the genes and the active compounds. [Results] A total of 314 RF-related DEGs were identified，with 202 genes upregulated and 112 genes downregulated. WGCNA identified the blue module as most relevant to RF，yielding 180 intersecting genes. Enrichment analysis revealed that immune responses and cytokine- mediated signaling pathways are critical in RF pathogenesis. Machine learning models identified five key feature genes：SERPINA3，CXCL10，JCHAIN，LTF，and CCL19. Immune infiltration analysis showed significant differences in immune cell infiltration between RF and control groups，and the feature genes correlated with immune cells. A total of 27 potential herbs were predicted，primarily targeting blood circulation，blood stasis，and deficiency， aligning with the TCM understanding of RF as a condition characterized by“deficiency and blood stasis”. Among the predicted herbs，Salvia miltiorrhiza，Curcumae rhizoma，Pheretima，and Conioselinum anthriscoides promote blood circulation，while Trionycis carapax，Ophiopogonis radix，Dendrobii caulis，and Ganoderma are tonifying，corresponding to RF’s pathogenesis. Molecular docking demonstrated strong binding between active TCM compounds and the key feature genes of RF. [Conclusion] SERPINA3，CXCL10，JCHAIN，LTF，and CCL19 are potential key genes for RF， and Salvia miltiorrhiza，Curcumae rhizoma，Pheretima，Conioselinum anthriscoides，Trionycis carapax， Ophiopogonis radix，Dendrobii caulis，and Ganoderma are potential TCM herbs for the treatment of RF. |
|
|
|
|
|
