Identification and validation of lactylation-related diagnostic biomarkers for type 2 diabetes by WGCNA

Figures

• 

  Fig. 1.. Differential expression and enrichment analyses. (A) Volcano plot of DEGs between normal and T2D samples in the GSE164416 training set. (B) Wikipathways enrichment analysis. (C) GO enrichment analysis. (D) KEGG pathway enrichment analysis.

• 

  Fig. 2.. Construction of WGCNA co-expression network and screening of key genes. (A) Soft threshold power screening plot. (B) Gene clustering dendrogram based on TOM. (C) Heatmap showing correlations between modules and T2D phenotypes, with numbers in squares indicating Pearson correlation coefficients and corresponding p values. (D) Venn diagram displaying overlaps among DEGs, module genes, and lactylation-related genes.

• 

  Fig. 3.. Identification of signature genes in T2D and evaluation of their diagnostic value. (A) Identification of signature genes through LASSO analysis. (B) Identification of signature genes using SVM-RFE analysis. (C) Identification of signature genes by RF analysis, with the figure showing the feature importance ranking. (D) Venn diagram illustrating the overlap of signature genes identified by the three machine learning algorithms. (E) Comparison of expression levels of consensus genes (identified by all three algorithms) between control and T2D groups. (F) ROC curves of diagnostic genes in the training set. (G) ROC curves of diagnostic genes in the validation set. (H) The mRNA levels of IKZF1, S100A4, and VIM in the cells were detected by qRT-PCR. (I) The protein levels of IKZF1, S100A4, and VIM in the cells were detected by WB. ***p<0.001, ****p<0.0001.

• 

  Fig. 4.. Functional pathway enrichment characteristics of diagnostic genes based on GSVA analysis. (A, B) GO (A) and KEGG (B) enrichment analyses of the diagnostic gene IKZF1. (C, D) GO (C) and KEGG (D) enrichment analyses of the diagnostic gene S100A4. (E, F) GO (E) and KEGG (F) enrichment analyses of the diagnostic gene VIM.

• 

  Fig. 5.. Analysis of immune infiltration characteristics and correlation with diagnostic genes. (A) Boxplots comparing immune cell and functional infiltration between T2D and control groups using ssGSEA algorithm. (B) Correlation analysis between IKZF1 and infiltrating immune cells/functions in T2D. (C) Correlation analysis between S100A4 and infiltrating immune cells/functions in T2D. (D) Correlation analysis between VIM and infiltrating immune cells/functions in T2D. ns, not significant; p>0.05; *p<0.05, **p<0.01, ***p<0.001.

• 

  Fig. 6.. Prediction analysis of miRNA and TFs regulatory networks. (A) PPI network of miRNAs with IKZF1, S100A4, and VIM. (B) PPI network of TFs with IKZF1, S100A4, and VIM. Red circles represent diagnostic genes. Blue squares represent miRNAs. Blue circles represent TFs.

• 

  Fig. 7.. Knocking down IKZF1, S100A4, and VIM can reduce their lactate levels and increase the viability of pancreatic β cells. (A) Transfection efficiency assessment using qRT-PCR. (B) Detection of the total protein lactate levels in each group by WB. (C) Measurement of the cell survival rate in each group using the CCK-8 assay. ****p<0.0001.