Mechanistic study of deoxycholic acid in colorectal cancer based on network toxicology and machine learning approaches.
[OBJECTIVE] This study aims to explore the mechanistic role of deoxycholic acid (DCA) in colorectal cancer (CRC) through the integration of network toxicology and machine learning approaches.
APA
Yin Y, Li X, et al. (2026). Mechanistic study of deoxycholic acid in colorectal cancer based on network toxicology and machine learning approaches.. BMC pharmacology & toxicology, 27(1), 32. https://doi.org/10.1186/s40360-026-01091-6
MLA
Yin Y, et al.. "Mechanistic study of deoxycholic acid in colorectal cancer based on network toxicology and machine learning approaches.." BMC pharmacology & toxicology, vol. 27, no. 1, 2026, pp. 32.
PMID
41559750
Abstract
[OBJECTIVE] This study aims to explore the mechanistic role of deoxycholic acid (DCA) in colorectal cancer (CRC) through the integration of network toxicology and machine learning approaches.
[METHODS] Gene differential expression analysis was first performed using datasets from the GEO database, comparing CRC and normal groups. Weighted Correlation Network Analysis (WCGNA) was subsequently applied to identify CRC-related genes. Target genes associated with DCA were predicted using the SEA and SwissTargetPrediction databases. The intersection of CRC-related genes and DCA-associated genes was then subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Further, 113 machine learning models were constructed, and the optimal model was selected based on the average AUC values of the training and validation sets. SHapley Additive exPlanations (SHAP) analysis was conducted to identify the top five genes contributing most to the model. These genes were further investigated for molecular docking with DCA.
[RESULTS] Differential gene expression analysis of the GEO dataset identified 371 CRC-related genes. WCGNA identified 5043 genes associated with CRC, and the union of these sets resulted in 5197 CRC-related genes. Using the SEA and SwissTargetPrediction databases, 55 and 59 DCA target genes were predicted, respectively, with the intersection yielding 102 DCA-associated genes. The intersection of CRC-related and DCA-associated genes resulted in 28 common genes, which were enriched in pathways such as monocarboxylate transport, steroid biosynthesis, and bile acid synthesis. Machine learning models identified EPHA4, G6PD, NR3C2, GPBAR1, and CRYAB as the top five most contributory genes. Molecular docking revealed that EPHA4 showed a strong binding affinity with DCA, particularly in the yellow and white regions of the protein, where amino acids like W45, K179, and D132 stabilized the ligand-protein interaction via hydrogen bonding and hydrophobic interactions.
[CONCLUSION] Deoxycholic acid promotes CRC progression through key genes such as EPHA4, G6PD, NR3C2, GPBAR1, and CRYAB. The EPH system may serve as a critical target for preventing DCA-induced CRC.
[CLINICAL TRIAL NUMBER] Not applicable.
[SUPPLEMENTARY INFORMATION] The online version contains supplementary material available at 10.1186/s40360-026-01091-6.
[METHODS] Gene differential expression analysis was first performed using datasets from the GEO database, comparing CRC and normal groups. Weighted Correlation Network Analysis (WCGNA) was subsequently applied to identify CRC-related genes. Target genes associated with DCA were predicted using the SEA and SwissTargetPrediction databases. The intersection of CRC-related genes and DCA-associated genes was then subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Further, 113 machine learning models were constructed, and the optimal model was selected based on the average AUC values of the training and validation sets. SHapley Additive exPlanations (SHAP) analysis was conducted to identify the top five genes contributing most to the model. These genes were further investigated for molecular docking with DCA.
[RESULTS] Differential gene expression analysis of the GEO dataset identified 371 CRC-related genes. WCGNA identified 5043 genes associated with CRC, and the union of these sets resulted in 5197 CRC-related genes. Using the SEA and SwissTargetPrediction databases, 55 and 59 DCA target genes were predicted, respectively, with the intersection yielding 102 DCA-associated genes. The intersection of CRC-related and DCA-associated genes resulted in 28 common genes, which were enriched in pathways such as monocarboxylate transport, steroid biosynthesis, and bile acid synthesis. Machine learning models identified EPHA4, G6PD, NR3C2, GPBAR1, and CRYAB as the top five most contributory genes. Molecular docking revealed that EPHA4 showed a strong binding affinity with DCA, particularly in the yellow and white regions of the protein, where amino acids like W45, K179, and D132 stabilized the ligand-protein interaction via hydrogen bonding and hydrophobic interactions.
[CONCLUSION] Deoxycholic acid promotes CRC progression through key genes such as EPHA4, G6PD, NR3C2, GPBAR1, and CRYAB. The EPH system may serve as a critical target for preventing DCA-induced CRC.
[CLINICAL TRIAL NUMBER] Not applicable.
[SUPPLEMENTARY INFORMATION] The online version contains supplementary material available at 10.1186/s40360-026-01091-6.
같은 제1저자의 인용 많은 논문 (5)
- CO transoral laser microsurgery for early glottic carcinoma with anterior commissure involvement.
- Reply: Persistent Racial Disparities in Lung Cancer Survival and the Overlooked Role of Post-Treatment Care.
- Gene Expression-Based Colorectal Cancer Prediction Using Machine Learning and SHAP Analysis.
- Racial/Ethnic Disparities in Non-Small Cell Lung Cancer Mortality in the U.S., 2000-2020: A Population-Based Study.
- Ultrasound Responsive Mn/Se-Nanozyme as PANoptosis Initiators for Bladder Cancer Immunotherapy.