Screening of immune-related secretory proteins linking chronic kidney disease with calcific aortic valve disease based on comprehensive bioinformatics analysis and machine learning

Figures

• 

  Fig. 1. Flow chart of this study design

• 

  Fig. 2. The integration of CAVD datasets and differential expression analysis of the integrated CAVD dataset. A PCA of three original CAVD datasets before batch-effect correction. B PCA of the integrated CAVD dataset after batch-effect correction. C The volcano plot representing CAVD DEGs in the integrated CAVD dataset. The upregulated genes are presented in red dots, whereas, the downregulated genes are presented in blue dots. D The heatmap showing the top 30 upregulated and 30 downregulated DEGs in the integrated CAVD dataset. CAVD calcific aortic valve disease, PCA principal component analysis, DEGs differentially expressed genes

• 

  Fig. 3. Screening of key module genes in the integrated CAVD dataset via WGCNA and identification of CAVD key genes through the intersection of key module genes and DEGs. A The scale-free topology model was utilized to identify the best β value, and β = 5 was chosen as the soft threshold based on the average connectivity and scale independence. B The network heatmap showing the gene dendrogram and module eigengenes. C The cluster dendrogram presenting module eigengenes. D The heatmap revealing the relationship between module eigengenes and status of CAVD. The correlation (upper) and p-value (bottom) of module eigengenes and status of CAVD were presented. The pink and yellow modules correlated with CAVD exhibited the highest and lowest correlation coefficients, respectively, which were identified as the key modules in CAVD. E The correlation plot between the pink module membership and the gene significance of genes in the pink module. F The correlation plot between the yellow module membership and the gene significance of genes in the yellow module. G A total of 124 key genes in CAVD were identified by taking the intersection between key modules genes and DEGs via the venn diagram. WGCNA…

• 

  Fig. 4. Identification of CKD-associated secretory proteins through differential expression analysis on secretory proteins in PMBC and kidney tissues of CKD. A The volcano plot revealing DEGs in the CKD-PMBC dataset. B The heatmap representing the top 30 upregulated and 30 downregulated DEGs in the CKD-PBMC dataset. C The volcano plot displaying DEGs in the CKD-Kidney dataset. D The heatmap displaying the top 30 upregulated and 30 downregulated DEGs in the CKD-Kidney dataset. E The intersection of CKD-PMBC DEGs with genes coding secretory proteins via the venn diagram. F The intersection of CKD-Kidney DEGs with genes coding secretory proteins via the venn diagram, and altogether 983 CKD-associated secretory proteins were identified. CKD chronic kidney disease, PBMC peripheral blood mononuclear cells, DEG differentially expressed genes

• 

  Fig. 5. PPI analysis between CKD-associated secretory proteins and CAVD key genes, and followed by enrichment analysis of the PPI-screened nodes. A The PPI network of module1 genes with the top1 highest score based on Cytoscape plug-in MCODE analysis. Salmon nodes are marked as members of CAVD key genes, yellow nodes as members of CKD-associated secretory proteins, while red nodes as common genes of the two sets. B The PPI network of module2 genes with the top2 highest score according to MCODE analysis. C–F The bubble plots showing the GO enrichment analysis results, including biological process (C), cellular component (D), and molecular function (E) of genes included in module1 and module2. F Circos plot representing the KEGG analysis results of genes included in module1 and module2. PPI protein-protein interaction, CKD chronic kidney disease, CAVD calcific aortic valve disease, MCODE molecular complex detection

• 

  Fig. 6. Screening of the potential small-molecular compounds for the treatment of CAVD via cMAP analysis. A The heatmap presenting the top10 compounds with the most significantly negative enrichment scores in 10 cell lines based on cMAP analysis. B The description of those top10 compounds. C The chemical structures of those 10 compounds were shown. cMAP connectivity map

• 

  Fig. 7. Identification of potential diagnostic biomarkers for CKD-related CAVD by the machine learning methods. A The venn diagram showing the 17 overlapping genes of CKD-associated secretory proteins, CAVD DEGs and CAVD key modules genes. B, C The minimum (B) and lambda values (C) of diagnostic biomarkers (n = 8) were identified by the LASSO logistic regression algorithm. D The RF algorithm presenting the MeanDecreaseGini of the 17 genes in CAVD and 6 biomarkers with the score more than 2.0 were selected. E The venn diagram displaying two common genes between LASSO and RF algorithms, which were identified as the hub genes in CKD-related CAVD. CKD chronic kidney disease, CAVD calcific aortic valve disease, DEGs differentially expressed genes, LASSO least absolute shrinkage and selection operator, RF random forest

• 

  Fig. 8. Development of the diagnostic nomogram model and efficacy assessment. A The nomogram was constructed based on the diagnostic biomarkers. B–D The ROC curve for the diagnostic performance of each candidate biomarker including SLPI (B), and MMP9 (C) and the nomogram model (D) constructed for CKD-related CAVD. E The calibration curve of nomogram model prediction in CKD-related CAVD. The dash line is marked as “Ideal”, which represents the standard curve, and is on behalf of the perfect prediction of the ideal model. The dotted line is marked as “Apparent”, which indicates the uncalibrated prediction curve, while the solid line is marked as “Bias-corrected” and represents the calibrated prediction curve. F DCA for the nomogram model. The black line is marked as “None”, which stands for the net benefit of the assumption that no patients have CAVD. The grey line is marked as “All”, which indicates the net benefit of the assumption that all patients have CAVD, and the purple line is marked as “Nomogram”, and represents the net benefit of the assumption that CKD-related CAVD are identified according to the diagnostic value of CAVD predicted by the nomogram model. G The ROC curve for the…

• 

  Fig. 9. Immune cell infiltration analysis in CAVD. A Stacked histogram displaying the immune cell proportions between CAVD and control groups. B Violin plot showing the comparison of 22 kinds of immune cells between CAVD and control groups. Red and blue stars represent the increased and decreased proportions of immune cells in CAVD group, respectively. C The heatmap revealing the correlation of 22 kinds of immune cells infiltration upon the threshold of p < 0.05. D The correlation map representing the association of the differentially infiltrated immune cells with two hub genes upon the threshold of p < 0.05. CAVD calcific aortic valve disease. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns not significant

• 

  Fig. 10. Validation of the expression patterns of two hub genes in calcified aortic valve samples and evaluation of the diagnostic performance of nomogram models in distinguishing CAVD. A RT-qPCR showing increased mRNA levels of SLPI and MMP9 in calcified aortic valve samples. B ELISA analysis displaying elevated serum SLPI and MMP9 levels in CKD and CAVD patients as well as CKD patients with CAVD. C The nomogram A was developed based on the diagnostic biomarkers to predict the risk of CAVD. D The ROC curves for the predictive performance of each candidate biomarker (SLPI and MMP9) and nomogram A. E The calibration curve of nomogram A prediction in CAVD patients. F DCA for the nomogram A. G The nomogram B was developed based on the diagnostic biomarkers to predict the risk of CKD patients with CAVD. H The ROC curves for the predictive performance of each candidate biomarker (SLPI and MMP9) and nomogram B. I The calibration curve of nomogram B prediction in CKD patients with CAVD. J DCA for the nomogram B