WGCNA算法在CKD基因模块构建中的应用

We used the WGCNA algorithm to construct gene modules closely related to sex, GFP, TIF, and CKD staging. To ensure the integrity of transcriptional information as much as possible, we performed WGCNA analysis using whole-genome information. As shown in Figure 2A, genes in cohort 1 were primarily divided into 10 modules. Among these, module brown was positively correlated with GFP and negatively correlated with TIF ratio and CKD staging; conversely, module blue and module turquoise were positively correlated with TIF ratio and CKD staging and negatively correlated with GFP. However, no modules were significantly associated with sex. Combining this with Figure 2B, module brown was the most significant module negatively correlated with CKD (r = -0.507, P = 0.014), while module blue showed the most significant positive correlation with CKD (TIF: r = 0.57, P = 0.0045; CKD staging: r = 0.653, P = 0.0068). Supplementary Figure 2A shows the correlations between genes in module blue and module brown and the traits. In Supplementary Figure 1A, one sample, S5, was identified as an outlier during sample clustering. Considering the authenticity and scarcity of the sample, we included it in the analysis. Supplementary Figures 1C and 1D show the soft power detection curves for cohort 1 (soft power = 16) and cohort 2 (soft power = 18), respectively, as well as the sample abundance distribution in Supplementary Figure 1C, indicating that the data quality of S5 was acceptable. WGCNA mainly divided the entire genome spectrum of cohort 2 into 5 modules. As shown in Figures 2C and 2D, module turquoise had the most significant positive correlation with CKD progression (r = 0.701, P = 0.0017), while module blue had the most significant negative correlation (r = -0.532, P = 0.028). Supplementary Figure 2B shows the correlations between genes in module turquoise and the traits. Similarly, no modules were found to be significantly associated with sex. In Supplementary Figure 1B, no outliers were detected in the sample clustering of cohort 2. As shown in Figures 2E and 2F, we intersected the differentially expressed genes from the module blue in cohort 1, which was significantly positively correlated with CKD, and the module turquoise in cohort 2, which was significantly positively correlated with CKD progression. We identified 9 genes in dataset GSE137570 that were significantly positively correlated with CKD development, naming them the positive gene set; similarly, a negative gene set consisting of 20 genes was significantly negatively correlated with CKD development. Through WGCNA, we initially screened out a risk gene set for CKD and a protective gene set for CKD.