2.6. Transcriptomic analysis

The gene expression patterns of IBD patients were analyzed using the weighted gene co-expression network analysis (WGCNA) method. WGCNA is a method for analyzing gene co-expression patterns. This method assumes that gene expression networks in biological systems follow a scale-free topology, which is characterized by a few hub genes with extensive connections and most other genes with relatively few connections. In these types of networks, the number of gene connections follows a negative exponential function with respect to their probability of occurrence. When using WGCNA, Pearson correlation coefficients between expression levels of all gene pairs are calculated first. Next, an appropriate soft threshold is selected to transform the correlation matrix into an adjacency matrix, which represents the connection strength between genes. Then, the topological overlap matrix (TOM) is constructed to measure the interconnectedness of genes. Finally, hierarchical clustering is performed on the basis of the TOM to group genes in modules of co-expressed genes. These steps enable the identification of gene modules with similar expression patterns and facilitate the discovery of gene interactions and potential biological functions. Genes selected via LASSO are then compared with the hub genes identified via WGCNA. If a selected gene is present in a specific WGCNA gene expression module, then all genes within that corresponding module will be extracted for further analysis [15].

Moreover, differentially expressed gene (DEG) analysis was conducted between the disease group/control group and between the high-score/low-score groups on the basis of LASSO results [11] via the Limma R package [56]. Through DEG analysis, we identified genes that were differentially expressed in respective groups. Due to the cascading amplification effect in gene expression, some genes with relatively small log2FC values may have significant impacts. Therefore, DEG analysis results are then filtered based only on P-adj<0.10 without applying a log2FC threshold [56].

The intersection of genes identified from the DEG analysis was examined and the union of these intersecting genes with genes identified from the WGCNA was determined. These intersecting genes were further subjected to gene enrichment analyses, including Gene Ontology (GO) analysis [14], Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis [13], and Gene Set Enrichment Analysis (GSEA) [16], using Hallmark gene sets. These analyses provided insights into the functional and pathway characteristics of genes.

Additionally, immune infiltration differences between the high- and low-LASSO score groups were examined via CIBERSORT [12], which is an approach to characterize the immune cell composition of complex tissues from their gene expression profiles. This analysis explored the differences in immune functions between the two groups and enhanced the understanding of their roles in IBD.