Table 1.
Primer sequences of Bmx and β-actin.
Fig 1.
The overall framework of this study.
Fig 2.
Identification of different genes and pathways between SCI and HC group.
(A) Boxplots of the GEO dataset distribution before and after normalization. (B) Volcano plot of DEGs, where red indicates upregulated DEGs, grey represents genes with no significant difference, and blue indicates downregulated DEGs. (C) Ridgeline plots display immune-related pathways from GSEA. (D) GSVA analysis of upregulated and downregulated pathways. (E) Heatmap showing the pathways analyzed by GSVA in SCI and HC samples.
Fig 3.
Identification of PANoptosis-related genes in SCI.
(A) The intersections between DEGs and PANoptosis-related genes. (B) Chromosomal positions of the 23 genes. (C) Heatmap showing the expression patterns of 23 genes between SCI and HC samples. (D) The scores of PANoptosis regulators between SCI samples and HC samples were compared.
Fig 4.
Functional analysis of the 23 common genes was performed.
(A) The detailed relationship between hub genes and the top 10 pathways annotated by GO functional enrichment analysis. (B) KEGG pathway enrichment analysis identified the top 10 pathways enriched in hub genes.
Fig 5.
The PPI network and the interactions among hub genes.
(A) A PPI network was constructed for the 23 common genes. (B) Using five algorithms (BottleNeck, Closeness, Degree, Betweenness, Stress) from the cytoHubba plugin, 10 common hub genes were identified. (C) Gene relationship circle diagram for the 10 hub genes, with red and blue lines representing positive and negative correlations, respectively. (D) The relationship between hub genes, with correlation coefficients indicated by the area of the pie chart. (E and F) Scatterplots were used to display the highest correlations among 10 hub genes: GADD45A and BMX (positive correlation), and TLR3 and GADD45A (negative correlation).
Fig 6.
Machine learning analysis of 10 hub genes related to SCI was performed using eight algorithms applied to the dataset.
The results from different machine learning methods include: Lasso-Logistic regression (A), LQV algorithm (B), Boruta (C), Bagged Tree (D), Random Forest (E), Bayesian (F), SVM (G) and xgboost (H). (I) A summary of the genes identified as important for SCI by all eight algorithms.
Fig 7.
Immune microenvironment analysis in the SCI and HC groups.
(A) The relative abundance of infiltrated immune cells between SCI and HC. (B) Pearson correlation matrix of these immune cells. (C) Pearson correlation analysis of the relationship between two hub genes and immune infiltration. (D) A lollipop plot was used to visualize the relationship between BMX and immune cells. (E) Scatterplot was used to display the correlation between BMX and Mast cells activated.
Fig 8.
Expression analysis (A) and ROC curve analysis of BMX (B) were performed.
Fig 9.
Pathway and functional analysis of the hub gene.
(A) The top five pathways positively associated with BMX. (B) The top five pathways negatively associated with BMX.
Fig 10.
Drug-gene, RBP-mRNA, and TF-mRNA network construction.
(A) Candidate drug molecules targeting the feature gene. (B) Candidate RBP targeting the feature gene. (C) Candidate TF targeting the feature gene.
Fig 11.
Experimental verification of BMX expression in SCI.
(A) Relative mRNA expression of Bmx in the spinal cord tissue of SCI and sham rat was evaluated by RT-qPCR. (B-C) WB analysis was used to detect BMX expression in spinal cord tissues, followed by densitometric quantification. (D) Images of IHC staining showing BMX expression in the Sham and SCI groups (scale bar = 50 μm). *P < 0.05, **P < 0.01, ***P < 0.001.