Expression patterns of pyroptosis-related genes in atrial fibrillation and their application in diagnostic models

Figures

• 

  Figure 1.. Flow outlining the PRDEG analysis approach.

• 

  Figure 2.. The removal of batch effects from GSE2240, GSE14975, and GSE41177. (A, B) The distribution boxplots for the combined dataset before (A) and after (B) removing batch effects. (C). PCA plots for the combined dataset before (C) and after (D) batch processing. The atrial fibrillation (AF) dataset GSE2240 is blue, the AF dataset GSE14975 is pink, and the AF dataset GSE41177 is purple. AF = atrial fibrillation; PCA = principal component analysis.

• 

  Figure 3.. Differential gene expression analyses. (A) Volcano plots for the DEGs identified when comparing the AF and control groups in the combined dataset. (B) Venn diagram showing the overlap between DEGs and PRGs in the combined dataset. (C) Heatmap of the top 20 PRDEGs in the combined dataset. (D) Chromosomal locations of PRDEGs. (E) Correlation heatmaps for 21 PRDEGs in the combined dataset. (F) Group comparison plots for PRDEGs in AF and control samples from the combined dataset. AF and control samples are, respectively, shown in pink and blue. In heatmaps, red and blue, respectively, represent high and low levels of expression. AF = atrial fibrillation; DEGs = differentially expressed genes; PRGs = pyroptosis-related genes; PRDEGs = pyroptosis-related differentially expressed genes.

• 

  Figure 4.. Functional enrichment analyses of PRDEGs. (A). Bubble plots for GO and KEGG pathway enrichment analyses for PRDEGs showing the biological process (BP), cell component (CC), and molecular function (MF) terms, as well as KEGG pathways. (B–E). GO and KEGG enrichment analyses for PRDEGs showing the BP (B), CC (C), MF (D), and KEGG (E) pathways. Items and molecules are shown in pink and blue, while lines represent associations between the 2. In bubble plots, bubble sizes are proportional to gene numbers, while pinker and bluer coloration, respectively, correspond to smaller and larger adjusted p-values. GO and KEGG enrichment results were screened based on a Benjamin–Hochberg adjusted P-value < .05 and an FDR (q-value) < 0.25. BP = biological process; CC = cellular component; FDR = false discovery rate; GO = Gene Ontology; KEGG = Kyoto Encyclopedia of Genes and Genomes; MF = molecular function; PRDEGs = pyroptosis-related differentially expressed genes.

• 

  Figure 5.. KEGG visualization. (A–C) KEGG pathway enrichment results were visualized for PRDEGs in the form of pathway maps for hepatitis B (A), human papillomavirus infection (B), and hepatitis C (C) pathways. KEGG = Kyoto Encyclopedia of Genes and Genomes; PRDEGs = pyroptosis-related differentially expressed genes.

• 

  Figure 6.. GSEA Analysis of the Combined Dataset. (A) GSEA analysis results showing 5 biological function bubbles for the combined GEO dataset. (B–F) GSEA results showing that all genes were enriched in the Wnt signaling (B), Focal Adhesion PI3KAKTM signaling (C), and IL-8 Cxcr2 (D). Glycolysis and gluconeogenesis (E), and aerobic glycolysis (F) pathways. In bubble plots, bubble sizes are proportional to gene numbers, while pinker and bluer coloration, respectively, correspond to smaller and larger adjusted P-values. GSEA results were screened based on a Benjamin–Hochberg adjusted P-value < .05 and an FDR (q-value) < 0.25. FDR = false discovery rate; GSEA = gene set enrichment analysis.

• 

  Figure 7.. GSVA analysis of the combined dataset. (A, B). Heatmaps (A) and group comparison maps (B) for GSVA results comparing the AF and control groups in the combined GEO dataset. Pink and blue samples, respectively, correspond to AF and control samples. P < .05 was the screening criterion for GSVA analyses. Blue and red, respectively, represent low and high levels of enrichment in heatmaps. AF = atrial fibrillation; GSVA = gene set variation analysis. ns = not significant; *P < .05, **P < .01, ***P < .001.

• 

  Figure 8.. Development of an AF diagnostic model. (A) Forest plots for 21 PRDEGs included in the logistic regression model used for the diagnosis of AF. (B, C). Numbers of genes with the lowest error rate (B) and the highest accuracy (C) derived from the SVM algorithm. (D, E). Diagnostic model plots (D) and variable trajectory plots (E) for the LASSO regression model. AF = atrial fibrillation; LASSO = Least Absolute Shrinkage and Selection Operator; PRDEGs = pyroptosis-related differentially expressed genes; SVM = support vector machine.

• 

  Figure 9.. Validation of the diagnostic model for AF. (A) Nomogram for the model genes in the AF diagnostic model in the combined GEO dataset. (B, C) Calibration curve (B) and decision curve analysis (C) plots for the model genes in the integrated GEO dataset when testing the AF diagnostic model. (D) ROC curves for RiskScore values in the combined GEO dataset. (E, F) ROC curves for the model genes CRTAC1, PKM, TLR3 (E), PTX3, SLC4A2, and HTRA1 (F) in AF samples. For calibration curves, the threshold probability is graphed against net benefit. An AUC > 0.5 was indicative of a trend towards a given gene contributing to the corresponding outcome, with an AUC closer to 1 being indicative of a stronger diagnostic effect, and values in the 0.7 to 0.9 range are indicative of some degree of accuracy. AF = atrial fibrillation; AUC = area under the curve; DCA = decision curve analysis; FPR = false positive rate; ROC = receiver operating characteristic; TPR = true positive rate.

• 

  Figure 10.. Differential gene expression and GSEA analyses for different risk groups. (A) Volcano plots for DEGs in the low- and high-risk AF samples from the combined GEO dataset. (B) The top 20 DEGs in the low- and high-risk AF samples. (C) GSEA analysis results for 5 biological function bubble plots corresponding to AF samples. (D–H) GSEA analyses showing that high-risk AF samples revealed significant enrichment in the PI3K Events In ERBB2 Signaling (D), IL-3 (E), P38 Mapk Signaling (F), TGF-beta Receptor Signaling In Skeletal Dysplasias (G), and glycolysis and gluconeogenesis (H) pathways. High- and low-risk samples are respectively shown in purple and green. In heatmaps, red and blue represent high and low levels of expression, respectively. In bubble plots, bubble sizes are proportional to gene numbers, while purple and green coloration respectively correspond to smaller and larger adjusted P-values. GSEA results were screened based on a Benjamin–Hochberg adjusted P < .05 and an FDR (q-value) < 0.25. AF = atrial fibrillation; FDR = false discovery rate; GSEA = Gene Set Enrichment Analysis.

• 

  Figure 11.. GSVA analyses for different risk groups. (A, B). Heat map (A) and group comparison map (B) results for the GSVA analysis of low- and high-risk AF samples from the combined GEO dataset. P < .05 was the cutoff for significance. *P < .05, **P < .01, ***P < .001. The high- and low-risk groups are respectively shown in purple and green. Low and high levels of enrichment are respectively shown in blue and red. GSVA = gene set variation analysis.

• 

  Figure 12.. PPI network analysis. (A) A PPI network was established with model genes developed using the STRING database. (B) An interaction network for functionally similar genes was developed with GeneMANIA. Circles correspond to model genes and genes with similar functions, while colors denote lines with interrelated functions. PPI = protein–protein interaction.