Using machine learning to find genes associated with sudden death

Figures

• 

  FIGURE 1. Differential expression and core gene screening. (A) Volcano plot of all genes. (Red dots represent up-regulated genes and green dots represent down-regulated genes). (B) Heatmap of differential genes in AD and SD groups. (Rows represent 47 differential genes and columns represent samples). (C) PPI network of differential genes. (Nodes represent hub genes. Lines represent interactions between hub genes). (D) Bar graph of all hub genes in the PPI network. (The x-axis represents channel counts. The y-axis represents hub genes). AD, Accidental death; SD, Sudden death.

• 

  FIGURE 2. Function, pathway, and disease enrichment analysis of core genes. (A) Bubble plot of GO function enrichment for core genes. (BP, Biological Process; CC, Cellular Components; MF, Molecular Function). (B) Circle plot of GO functional enrichment. (C) Bubble map of KEGG pathway enrichment for core genes. (D) Circle plot of KEGG enrichment analysis. (E) Bubble plot of DO enrichment for core genes. (F) Circle plot of DO enrichment analysis. The size of bubbles in the bubble plot represents the number of core genes in the corresponding pathway. The color of the bubbles represents the adjusted p-value. The circle plot illustrates certain core genes corresponding to the GO/KEGG terminology or disease. LogFC represents the expression level of gene.

• 

  FIGURE 3. Machine learning identifies characteristic genes of sudden death. (A) The LASSO regression algorithm was used to select the characteristic genes of sudden death. (B) SVM-RFE algorithm to select the characteristic genes of sudden death. (The blue point represents the lowest error rate, correspondingly to the best genome selected by SVM-RFE). (C) Venn diagram showing 2 sudden death characteristic genes shared by LASSO (green) and SVM-RFE (pink) algorithms. LASSO, least absolute shrinkage and selector operation. SVM-RFE, support vector machine-recursive feature elimination.

• 

  FIGURE 4. Expression and ROC curves of characteristic genes of sudden death. (A) The expression level of MYL2 in AD and SD groups. (B) The expression level of TNNT3 in AD and SD groups. (C) The ROC curve of MYL2. (D) The ROC curve of TNNT3. ROC, receiver operating characteristic; AD, Accidental death; SD, Sudden death.

• 

  FIGURE 5. Expression of sudden death characteristic genes in human tissues. (A,B) MYL2 and TNNT3 expression levels in tissues in males and females. (C) Validation of MYL2 in turquoise module by HPA (IHC). (D) Validation ofTNNT3 in turquoise module by HPA (IHC). Red represents high expression, green represents low expression, and black represents mediate expression.

• 

  FIGURE 6. Profile and visualization of immune cell infiltration. (A) The infiltration of 22 immune cells after quantification by the CIBERSORT algorithm. (The X-axis represents the sample and the Y-axis shows the percentage of 22 immune cells in the sample as stacked bars). (B) Violin plot showing comparison based on 22 immune cells. (Blue and red represent AD and SD group samples, respectively). AD, Accidental death; SD, Sudden death.