Table 1.
Primer sequences.
Fig 1.
Identification and functional analysis of candidate genes associated with mitochondrial dynamics in IPF.
(A) Volcano plot of differentially expressed genes between IPF and control groups. (B) Heatmap of differentially expressed genes. (C) Consensus clustering heatmap. (D) Volcano plot of differentially expressed genes between Cluster1 and Cluster2. (E) Venn diagram of candidate genes. (F) GO and KEGG enrichment analyses of candidate genes (the top three pathways). (G) PPI network of candidate genes. (H) Genes in the top-scoring module.
Fig 2.
Identification of CD247, RETN, and IL7R as biomarkers using a comprehensive analysis of 101 machine-learning models.
(A) AUC values of the models on the training and validation sets. (B) and (C) ROC curves of the optimal model in the training and validation sets, respectively. (D) and (E) Boxplots of feature gene expression levels in the training and validation sets, respectively. Red represents IPF samples, and blue represents control samples. (F) and (G) ROC curves of feature genes in the training and validation sets, respectively.
Table 2.
Performance metrics derived from the confusion matrices.
Fig 3.
Predictive performance of CD247, RETN, and IL7R in IPF.
(A) Architecture of the artificial neural network (ANN) model. Nodes represent neurons, and edges represent weighted connections; color intensity reflects connection strength. (B–C) Confusion matrices for the training and validation cohorts. (D–E) Receiver operating characteristic (ROC) curves for the training and validation cohorts. The ANN achieved an AUC of 0.91 in the training set and 0.82 in the validation set.
Fig 4.
Analysis of chromosomal localization and functional assessment of CD247, RETN, and IL7R.
(A) Chromosomal localization map. (B) Correlation heatmap of biomarkers. (C) GeneMANIA network of biomarkers. The circles in the figure represent co-expressed genes of the biomarkers, and different colored blocks reflect their involved functions. (D), (E), and (F) Top three GSEA-enriched pathways for CD247, RETN, and IL7R, respectively.
Fig 5.
Immune infiltration analysis of CD247, RETN, and IL7R.
(A) Heatmap of immune cell enrichment scores in the IPF and control groups. The color represents the enrichment score, with yellow indicating a score greater than 0 and blue representing a score less than 0. The darker the color, the higher the score. (B) Differences in immune cell enrichment scores between IPF patients and control samples. Yellow represents IPF samples, and blue represents control samples. (C) Correlation heatmap of differential immune cell populations. Blue represents negative correlation, yellow represents positive correlation, and the darker the color, the greater the correlation. Blank areas in the figure represent insignificant p-values. (D) Correlations between differential immune cell populations and biomarkers. Blue represents negative correlation, yellow represents positive correlation, and the darker the color, the greater the correlation. The size of the circle indicates whether the p-value is significant, with numbers on the circles representing the magnitude of the correlation.
Fig 6.
Regulatory network analysis of CD247, RETN, and IL7R.
(A) Transcription factor (TF)–mRNA regulatory network of the biomarkers. Green nodes represent biomarkers, and blue nodes represent TF. (B) lncRNA–miRNA–mRNA regulatory network of the biomarkers. Pink nodes represent genes, blue nodes represent miRNAs, and green nodes represent lncRNAs. (C) circRNA–miRNA–mRNA regulatory network of the biomarkers. Pink nodes represent genes, blue nodes represent miRNAs, and purple nodes represent circRNAs.
Table 3.
Drug prediction and molecular docking profiling.
Fig 7.
Drug prediction and molecular docking profiles.
Molecular docking diagrams for CD247 (A), IL7R (B), and RETN (C-D) are shown from top to bottom. The stick models represent the molecular structures of the active compounds, with labeled amino acid residues indicating the docking sites. The left panels display the overall docking views, while the right panels provide magnified local views.
Fig 8.
Annotation of four major immune cell types and identification of monocytes as the predominant biomarker-associated cell population.
(A) UMAP plot of cell clustering. (B) Cell clustering plot of IPF and control samples. (C) Cell type annotation. (D) Heatmap of functional enrichment analysis. Colors represent enrichment scores. (E) UMAP plot of biomarker expression levels. (F) Expression of biomarkers across different cell types. Cell types are ordered by the average expression values. (G) Violin plot of biomarker expression levels.
Fig 9.
Cellular communication and pseudotime trajectory analysis.
(A) Interaction count network between monocytes and other cell types. The nodes represent different cell types, and the size of the nodes indicates the number of cells in each type. The thickness of the lines represents the number of communications. (B) Interaction strength network between monocytes and other cell types. The nodes represent different cell types, and the size of the nodes indicates the number of cells in each type. The thickness of the lines represents the strength of communications. (C) Interaction diagram of ligand-receptor pairs between cell types. The vertical axis represents the signaling cells, and the horizontal axis represents the receiving cells. The color intensity of the heatmap indicates the strength of the signal. The bars on the top and right side represent the cumulative counts of the vertical and horizontal axes, respectively. (D) Pseudotime trajectory of cell differentiation. Dark blue represents the early stages of differentiation, while light blue represents the later stages. (E) Different states of cells. (F) Expression trends of biomarkers during cell differentiation.
Fig 10.
Validation of biomarker expression levels in peripheral blood samples from 10 patients with IPF and 10 healthy controls.
(A) The mRNA expression level of RETN. (B) The mRNA expression level of CD247. (C) The mRNA expression level of IL7R. Note: **p < 0.01, ***p < 0.001.