Issue: Vol.82 (No. 8)

Observational study on the potential mechanism of SiNi powder in the treatment of ulcerative colitis based on network pharmacology and machine learning

Authors:
Sihong Shen, Yongduo Yu

Download full articele PDF

Background/ Aim. Chronic idiopathic ulcerative colitis (UC) damages and disrupts the intestinal mucosa. Diagnosing UC and differential diagnosis is tough. Its anti-inflammatory and immunosuppressive properties make SiNi powder (SNP) a popular treatment for inflammatory illnesses. The multi-target mechanism of SNP on UC is unknown. The aim of this study was to examine the potential mechanisms of SNP in UC treatment through network pharmacology and machine learning approaches, identify novel diagnostic biomarkers, and develop a predictive model for UC diagnosis. Methods. The Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) assessed active constituents and target proteins. Using two public datasets (GSE87473 and GSE75214), differential analysis was conducted on the gene expression matrix of UC to find the intersection of differentially expressed genes and SNP-related targets. Hub genes were assessed using several machine-learning algorithms to create a prediction model. Single-cell analysis studies were used to diagnose genes and immune cells. NetworkAnalyst predicted upstream transcription factors, micro-ribonucleic acids, and the protein-compound network. Results. According to the TCMSP database, the SNP included 95 active constituents and 795 associated targets against UC. After identifying 79 overlapping genes, machine learning discovered five hub genes: TRPV1, ABCG2, BACE2, MMP3, and LIPC. Diagnostics were verified using external datasets. These genes were used to create a predictive model with a large area under the curve (AUC = 1,000) and an external validation dataset with 1,000 AUCs, demonstrating excellent accuracy of the predictive model and the hub genes. Conclusion. SNP and UC are associated, and hub genes were found to evaluate UC risk. This computational technique opens new avenues for UC biomarker and therapeutic target research, although further experimental validation is required to confirm and validate these results.