Machine Learning-Based Transcriptomic Diagnosis of Periodontitis

Exploration of the shared diagnostic genes and mechanisms between periodontitis and primary Sjögren’s syndrome by integrated comprehensive bioinformatics analysis and machine learning

Proliferative diabetic retinopathy (PDR) is a vision-threatening complication of diabetes, in which hypoxia plays a central pathogenic role. However, the hypoxia-associated molecular mechanisms and biomarkers in PDR remain incompletely understood. RNA sequencing data from patients with PDR and healthy controls (GSE146615) were analyzed to identify differentially expressed genes (DEGs). Hypoxia-related genes (HRGs) were obtained from the GeneCards database and integrated with DEGs and weighted gene co-expression network analysis (WGCNA) modules. Candidate genes were refined using least absolute shrinkage and selection operator (LASSO) regression and extreme gradient boosting (XGBoost). Diagnostic performance was assessed by receiver operating characteristic (ROC) analysis. Immune infiltration was estimated with the CIBERSORT algorithm, and biomarker–immune cell correlations were examined. We identified 1,650 DEGs in PDR, enriched in immune regulation, vascular function, and mitochondrial pathways. Intersection analysis identified 13 hypoxia-related genes, of which four—CXCL9, DSC2, DSC3, and PITRM1—were selected as key biomarkers by LASSO and XGBoost. ROC analysis showed strong diagnostic performance for PITRM1 (AUC = 0.863), DSC2 (AUC = 0.861), DSC3 (AUC = 0.837), and CXCL9 (AUC = 0.749). Immune infiltration analysis revealed increased plasma cells and CD8⁺ T cells, and decreased resting mast cells in PDR. This integrative bioinformatics analysis identified four hypoxia-related genes as potential diagnostic biomarkers for PDR, providing insights into hypoxia-driven immune and vascular changes in disease pathogenesis. These findings may inform future diagnostic and therapeutic strategies.

Background The genetic factors and pathogenesis of idiopathic dilated cardiomyopathy-induced heart failure (IDCM-HF) have not been understood thoroughly; there is a lack of specific diagnostic markers and treatment methods for the disease. Hence, we aimed to identify the mechanisms of action at the molecular level and potential molecular markers for this disease. Methods Gene expression profiles of IDCM-HF and non-heart failure (NF) specimens were acquired from the database of Gene Expression Omnibus (GEO). We then identified the differentially expressed genes (DEGs) and analyzed their functions and related pathways by using “Metascape”. Weighted gene co-expression network analysis (WGCNA) was utilized to search for key module genes. Candidate genes were identified by intersecting the key module genes identified via WGCNA with DEGs and further screened via the support vector machine-recursive feature elimination (SVM-RFE) method and the least absolute shrinkage and selection operator (LASSO) algorithm. At last, the biomarkers were validated and evaluated the diagnostic efficacy by the area under curve (AUC) value and further confirmed the differential expression in the IDCM-HF and NF groups using an external database. Results We detected 490 genes exhibiting differential expression between IDCM-HF and NF specimens from the GSE57338 dataset, with most of them being concentrated in the extracellular matrix (ECM) of cells related to biological processes and pathways. After screening, 13 candidate genes were identified. Aquaporin 3 (AQP3) and cytochrome P450 2J2 (CYP2J2) showed high diagnostic efficacy in the GSE57338 and GSE6406 datasets, respectively. In comparison to the NF group, AQP3 was significantly down-regulated in the IDCM-HF group, while CYP2J2 was significantly up-regulated. Conclusion As far as we know, this is the first study that combines WGCNA and machine learning algorithms to screen for potential biomarkers of IDCM-HF. Our findings suggest that AQP3 and CYP2J2 could be used as novel diagnostic markers and treatment targets of IDCM-HF.

Identification and validation of immune-related biomarkers and polarization types of macrophages in keloid based on bulk RNA-seq and single-cell RNA-seq analysis.

Esophageal squamous cell carcinoma (ESCC) ranks as the fourth leading cause of cancer-related death in China. Although paclitaxel has been shown to be effective in treating ESCC, the prolonged use of this chemical will lead to paclitaxel resistance. In order to uncover genes and pathways driving paclitaxel resistance in the progression of ESCC, bioinformatics analyses were performed based on The Cancer Genome Atlas (TCGA) database and the Gene Expression Omnibus (GEO) database including GSE86099 and GSE161533. Differential expression analysis was performed in TCGA data and two GEO datasets to obtain differentially expressed genes (DEGs). Based on GSE161533, weighted gene co-expression network analysis (WGCNA) was conducted to identify the key modules associated with ESCC tumor status. The DEGs common to the two GEO datasets and the genes in the key modules were intersected to obtain the paclitaxel resistance-specific or non-paclitaxel resistance-specific genes, which were subjected to subsequent least absolute shrinkage and selection operator (LASSO) feature selection, whereby paclitaxel resistance-specific or non-paclitaxel resistance-specific key genes were selected. Ten machine learning models were used to validate the biological significance of these key genes; the potential therapeutic drugs for paclitaxel resistance-specific genes were also predicted. As a result, we identified 24 paclitaxel resistance-specific genes and 18 non-paclitaxel resistance-specific genes. The ESCC machine classifiers based on the key genes achieved a relatively high AUC value in the cross-validation and in an independent test set, GSE164158. A total of 207 drugs (such as bevacizumab) were predicted to be alternative therapeutics for ESCC patients with paclitaxel resistance. These results might shed light on the in-depth research of paclitaxel resistance in the context of ESCC progression.

Background Diabetic nephropathy (DN), a significant cause of chronic kidney disease (CKD), is a devastating disease worldwide. Objective The aim of this study was to reveal crucial genes closely linked to the molecular mechanism of tubulointerstitial injury in DN. Methods The Gene Expression Omnibus (GEO) database was used to download the datasets. Based on this, a weighted gene coexpression network analysis (WGCNA) network was constructed to detect DN-related modules and hub genes. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichments were performed on the selected hub genes and modules. Least absolute shrinkage and selection operator (LASSO) Cox regression analysis was performed on the obtained gene signature. Results The WGCNA network was constructed based on 3019 genes, and nine gene coexpression modules were generated. A total of 57 genes, including 34 genes in the magenta module and 23 genes in the purple module, were adapted as hub genes. 61 significantly downregulated and 119 upregulated genes were screened as differentially expressed genes (DEGs). 25 overlapping genes between hub genes chosen from WGCNA and DEG were identified. Through LASSO analysis, a 9-gene signature may be a potential prognostic biomarker for DN. To further explore the potential mechanism of DN, the different immune cell infiltrations between tubulointerstitial samples of DN and healthy samples were estimated. Conclusions This bioinformatics study identified CX3CR1, HRG, LTF, TUBA1A, GADD45B, PDK4, CLIC5, NDNF, and SOCS2 as candidate biomarkers for the diagnosis of DN. Moreover, DN tends to own a higher proportion of memory B cell.

BackgroundTemporal lobe epilepsy (TLE) is a common chronic episodic illness of the nervous system. However, the precise mechanisms of dysfunction and diagnostic biomarkers in the acute phase of TLE are uncertain and hard to diagnose. Thus, we intended to qualify potential biomarkers in the acute phase of TLE for clinical diagnostics and therapeutic purposes.MethodsAn intra-hippocampal injection of kainic acid was used to induce an epileptic model in mice. First, with a TMT/iTRAQ quantitative labeling proteomics approach, we screened for differentially expressed proteins (DEPs) in the acute phase of TLE. Then, differentially expressed genes (DEGs) in the acute phase of TLE were identified by linear modeling on microarray data (limma) and weighted gene co-expression network analysis (WGCNA) using the publicly available microarray dataset GSE88992. Co-expressed genes (proteins) in the acute phase of TLE were identified by overlap analysis of DEPs and DEGs. The least absolute shrinkage and selection operator (LASSO) regression and support vector machine recursive feature elimination (SVM-RFE) algorithms were used to screen Hub genes in the acute phase of TLE, and logistic regression algorithms were applied to develop a novel diagnostic model for the acute phase of TLE, and the sensitivity of the diagnostic model was validated using receiver operating characteristic (ROC) curves.ResultsWe screened a total of 10 co-expressed genes (proteins) from TLE-associated DEGs and DEPs utilizing proteomic and transcriptome analysis. LASSO and SVM-RFE algorithms for machine learning were applied to identify three Hub genes: Ctla2a, Hapln2, and Pecam1. A logistic regression algorithm was applied to establish and validate a novel diagnostic model for the acute phase of TLE based on three Hub genes in the publicly accessible datasets GSE88992, GSE49030, and GSE79129.ConclusionOur study establishes a reliable model for screening and diagnosing the acute phase of TLE that provides a theoretical basis for adding diagnostic biomarkers for TLE acute phase genes.

To investigate the signature genes of fatty acid metabolism and their association with immune cells in pulmonary arterial hypertension (PAH). Fatty acid metabolism-related genes were obtained from the GeneCards database. In this retrospective study, a PAH-related dataset was downloaded from the Gene Expression Omnibus database and analyzed to identify differentially expressed genes (DEGs). Weighted gene co-expression network analysis (WGCNA) and machine learning algorithms, including least absolute shrinkage and selection operator (LASSO) and random forest, were used to identify the signature genes. Diagnostic efficiency was assessed by receiver operating characteristic (ROC) curve analysis and a nomogram. Immune cell infiltration was subsequently classified using CIBERSORT. In total, 817 DEGs were screened from the GSE33463 dataset. The data were clustered into six modules via WGCNA, and the MEdarkred module was significantly related to PAH. The LASSO and random forest algorithms identified five signature genes: ARV1, KCNJ2, PEX11B, PITPNC1, and SCO1. The areas under the ROC curves of these signature genes were 0.917, 0.934, 0.947, 0.963, and 0.940, respectively. CIBERSORT suggested these signature genes may participate in immune cell infiltration. ARV1, KCNJ2, PEX11B, PITPNC1, and SCO1 show remarkable diagnostic performance in PAH and are involved in immune cell infiltration.

BackgroundImmune system dysregulation plays a critical role in aortic valve calcification (AVC) and metabolic syndrome (MS) pathogenesis. The study aimed to identify pivotal diagnostic candidate genes for AVC patients with MS.MethodsWe obtained three AVC and one MS dataset from the gene expression omnibus (GEO) database. Identification of differentially expressed genes (DEGs) and module gene via Limma and weighted gene co-expression network analysis (WGCNA), functional enrichment analysis, protein–protein interaction (PPI) network construction, and machine learning algorithms (least absolute shrinkage and selection operator (LASSO) regression and random forest) were used to identify candidate immune-associated hub genes for diagnosing AVC with MS. To assess the diagnostic value, the nomogram and receiver operating characteristic (ROC) curve were developed. Finally, immune cell infiltration was created to investigate immune cell dysregulation in AVC.ResultsThe merged AVC dataset included 587 DEGs, and 1,438 module genes were screened out in MS. MS DEGs were primarily enriched in immune regulation. The intersection of DEGs for AVC and module genes for MS was 50, which were mainly enriched in the immune system as well. Following the development of the PPI network, 26 node genes were filtered, and five candidate hub genes were chosen for nomogram building and diagnostic value evaluation after machine learning. The nomogram and all five candidate hub genes had high diagnostic values (area under the curve from 0.732 to 0.982). Various dysregulated immune cells were observed as well.ConclusionFive immune-associated candidate hub genes (BEX2, SPRY2, CXCL16, ITGAL, and MORF4L2) were identified, and the nomogram was constructed for AVC with MS diagnosis. Our study could provide potential peripheral blood diagnostic candidate genes for AVC in MS patients.

Neuropathic pain (NP) is a debilitating disease and is associated with energy metabolism alterations. This study aimed to identify energy metabolism-related differentially expressed genes (EMRDEGs) in NP, construct a diagnostic model, and analyze immune cell infiltration and single-cell gene expression characteristics of NP. GSE89224, GSE123919, and GSE134003 were downloaded from the Gene Expression Omnibus. Differentially expressed genes (DEGs) analysis and an intersection with highly energy metabolism-related modules in weighted gene co-expression network analysis (WGCNA) was performed in GSE89224. Least absolute shrinkage and selection operator (LASSO), random forest, and logistic regression were used for model genes selection. NP samples were divided into high- and low-risk groups and different disease subtypes based on risk score of LASSO algorithm and consensus clustering analysis, respectively. Immune cell composition was estimated in different risk groups and NP subtypes. Datasets 134,003 were performed for identification of single-cell DEGs and functional enrichment. Cell-cell communications and pseudo-time analysis to reveal the expression profile of NP. A total of 38 EMRDEGs were obtained and are majorly enriched in metabolism about glioma and inflammation. LASSO, random forest, and logistic regression identified 6 model genes, which were Itpr1, Gng8, Socs3, Fscn1, Cckbr, and Camk1. The nomogram, based on six model genes, had a good predictive ability, concordance, and diagnostic value. The comparisons between different risk groups and NP subtypes identified important pathways and different immune cells component. The immune infiltration results majorly associated with inflammation and energy metabolism. Single-cell analysis revealed cell-cell communications and cells differentiation characteristics of NP. In conclusion, our results not only elucidate the involvement of energy metabolism in NP but also provides a robust diagnostic tool with six model genes. These findings might give insight into the pathogenesis of NP and provide effective therapeutic regimens for the treatment of NP.

Immunogenic cell death (ICD) is a form of programmed cell death with a strong sense of inflammatory detection, whose powerful situational awareness can cause the reactivation of aberrant immunity. However, the role of ICD in the pathogenesis of severe acute pancreatitis (SAP) has yet to be investigated. This study aims to explore the pivotal genes associated with ICD in SAP and how they relate to immune infiltration and short-chain fatty acids (SCFAs), in order to provide a theoretical foundation for further, in-depth mechanistic studies. We downloaded GSE194331 datasets from the Gene Expression Omnibus (GEO). The use of differentially expressed gene (DEG) analysis; weighted gene co-expression network analysis (WGCNA) and least absolute shrinkage and selection operator (LASSO) regression analysis allowed us to identify a total of three ICD-related hub genes (LY96, BCL2, IFNGR1) in SAP. Furthermore, single sample gene set enrichment analysis (ssGSEA) demonstrated that hub genes are closely associated with the infiltration of specific immune cells, the activation of immune pathways and the metabolism of SCFAs (especially butyrate). These findings were validated through the analysis of gene expression patterns in both clinical patients and rat animal models of SAP. In conclusion, the first concept of ICD in the pathogenesis of SAP was proposed in our study. This has important implications for future investigations into the pro-inflammatory immune mechanisms mediated by damage-associated molecular patterns (DAMPs) in the late stages of SAP.

BackgroundBipolar disorder and metabolic syndrome are both associated with the expression of immune disorders. The current study aims to find the effective diagnostic candidate genes for bipolar affective disorder with metabolic syndrome.MethodsA validation data set of bipolar disorder and metabolic syndrome was provided by the Gene Expression Omnibus (GEO) database. Differentially expressed genes (DEGs) were found utilizing the Limma package, followed by weighted gene co-expression network analysis (WGCNA). Further analyses were performed to identify the key immune-related center genes through function enrichment analysis, followed by machine learning-based techniques for the construction of protein–protein interaction (PPI) network and identification of the Least Absolute Shrinkage and Selection Operator (LASSO) and Random Forest (RF). The receiver operating characteristic (ROC) curve was plotted to diagnose bipolar affective disorder with metabolic syndrome. To investigate the immune cell imbalance in bipolar disorder, the infiltration of the immune cells was developed.ResultsThere were 2,289 DEGs in bipolar disorder, and 691 module genes in metabolic syndrome were identified. The DEGs of bipolar disorder and metabolic syndrome module genes crossed into 129 genes, so a total of 5 candidate genes were finally selected through machine learning. The ROC curve results-based assessment of the diagnostic value was done. These results suggest that these candidate genes have high diagnostic value.ConclusionPotential candidate genes for bipolar disorder with metabolic syndrome were found in 5 candidate genes (AP1G2, C1orf54, DMAC2L, RABEPK and ZFAND5), all of which have diagnostic significance.

ObjectivesPost-stroke depression (PSD) is a common and serious psychiatric complication which hinders functional recovery and social participation of stroke patients. Stroke is characterized by dynamic changes in metabolism and hemodynamics, however, there is still a lack of metabolism-associated effective and reliable diagnostic markers and therapeutic targets for PSD. Our study was dedicated to the discovery of metabolism related diagnostic and therapeutic biomarkers for PSD.MethodsExpression profiles of GSE140275, GSE122709, and GSE180470 were obtained from GEO database. Differentially expressed genes (DEGs) were detected in GSE140275 and GSE122709. Functional enrichment analysis was performed for DEGs in GSE140275. Weighted gene co-expression network analysis (WGCNA) was constructed in GSE122709 to identify key module genes. Moreover, correlation analysis was performed to obtain metabolism related genes. Interaction analysis of key module genes, metabolism related genes, and DEGs in GSE122709 was performed to obtain candidate hub genes. Two machine learning algorithms, least absolute shrinkage and selection operator (LASSO) and random forest, were used to identify signature genes. Expression of signature genes was validated in GSE140275, GSE122709, and GSE180470. Gene set enrichment analysis (GSEA) was applied on signature genes. Based on signature genes, a nomogram model was constructed in our PSD cohort (27 PSD patients vs. 54 controls). ROC curves were performed for the estimation of its diagnostic value. Finally, correlation analysis between expression of signature genes and several clinical traits was performed.ResultsFunctional enrichment analysis indicated that DEGs in GSE140275 enriched in metabolism pathway. A total of 8,188 metabolism associated genes were identified by correlation analysis. WGCNA analysis was constructed to obtain 3,471 key module genes. A total of 557 candidate hub genes were identified by interaction analysis. Furthermore, two signature genes (SDHD and FERMT3) were selected using LASSO and random forest analysis. GSEA analysis found that two signature genes had major roles in depression. Subsequently, PSD cohort was collected for constructing a PSD diagnosis. Nomogram model showed good reliability and validity. AUC values of receiver operating characteristic (ROC) curve of SDHD and FERMT3 were 0.896 and 0.964. ROC curves showed that two signature genes played a significant role in diagnosis of PSD. Correlation analysis found that SDHD (r = 0.653, P < 0.001) and FERM3 (r = 0.728, P < 0.001) were positively related to the Hamilton Depression Rating Scale 17-item (HAMD) score.ConclusionA total of 557 metabolism associated candidate hub genes were obtained by interaction with DEGs in GSE122709, key modules genes, and metabolism related genes. Based on machine learning algorithms, two signature genes (SDHD and FERMT3) were identified, they were proved to be valuable therapeutic and diagnostic biomarkers for PSD. Early diagnosis and prevention of PSD were made possible by our findings.

Background: Venous thromboembolism (VTE) is a major cause of morbidity and mortality globally, accompanied with a variety of underlying diseases, and it often doesn't get enough attention for early diagnosis and treatment. VTE is associated with many risk factors, however, the genetic basis of venous thromboembolism remains poorly understood. Methods: We analyzed whole-blood gene expression profiles from 185 VTE patients and 93 normal controls obtained from the GEO database. Differential expression analysis was conducted to identify differentially expressed genes (DEGs), followed by weighted gene co-expression network analysis (WGCNA) to identify key genes through intersection analysis. Subsequently, enrichment analysis was performed on gene modules associated with key genes. Central genes were identified using the Least Absolute Shrinkage and Selection Operator (LASSO) and random forest algorithms. Additionally, immune cell infiltration was analyzed using CIBERSORT, and pathway functions of key genes were assessed through Gene Set Enrichment Analysis (GSEA). Results: A total of 102 DEGs were identified in VTE. Using weighted gene co-expression network analysis (WGCNA), these DEGs were clustered into 14 modules. Among them, the black module showed significant correlation with VTE (cor=0.48, p&lt;0.0001). Machine learning techniques identified four key genes, namely TXNL1 (AUC=0.840), CD3G (AUC=0.786), MRPL15 (AUC=0.775), and MCTS1 (AUC=0.748), demonstrating high diagnostic efficacy for predicting VTE. These four hub genes were significantly associated with pathways such as TRANSLATION, SRP DEPENDENT COTRANSLATIONAL PROTEIN TARGETING TO MEMBRANE, and EUKARYOTIC TRANSLATION INITIATION, providing valuable insights into the potential mechanisms underlying VTE. Immunoinfiltration analysis using ssGSTEA revealed significant positive correlations between Activated CD8 T cell, Gamma delta T cell, and hub genes, suggesting the potential of these hub genes and immune cell populations for targeted immunotherapy in VTE patients, thereby opening new avenues for therapeutic intervention. Conclusion: TXNL1, CD3G, MRPL15, and MCTS1 have emerged as promising candidate driver genes for VTE. These central genes hold the potential to provide valuable insights into the molecular diagnosis and treatment of VTE.

Acute respiratory distress syndrome (ARDS), a progressive status of acute lung injury (ALI), is primarily caused by an immune-mediated inflammatory disorder, which can be an acute pulmonary complication of rheumatoid arthritis (RA). As a chronic inflammatory disease regulated by the immune system, RA is closely associated with the occurrence and progression of respiratory diseases. However, it remains elusive whether there are shared genes between the molecular mechanisms underlying RA and ARDS. The objective of this study is to identify potential shared genes for further clinical drug discovery through integrated analysis of bulk RNA sequencing datasets obtained from the Gene Expression Omnibus database, employing differentially expressed genes (DEGs) analysis and weighted gene co-expression network analysis (WGCNA). The hub genes were identified through the intersection of common DEGs and WGCNA-derived genes. The Random Forest (RF) and least absolute shrinkage and selection operator (LASSO) algorithms were subsequently employed to identify key shared target genes associated with two diseases. Additionally, RA immune infiltration analysis and COVID-19 single-cell transcriptome analysis revealed the correlation between these key genes and immune cells. A total of 59 shared genes were identified from the intersection of DEGs and gene clusters obtained through WGCNA, which analyzed the integrated gene matrix of ALI/ARDS and RA. The RF and LASSO algorithms were employed to screen for target genes specific to ALI/ARDS and RA, respectively. The final set of overlapping genes (FCMR, ADAM28, HK3, GRB10, UBE2J1, HPSE, DDX24, BATF, and CST7) all exhibited a strong predictive effect with an area under the curve (AUC) value greater than 0.8. Then, the immune infiltration analysis revealed a strong correlation between UBE2J1 and plasma cells in RA. Furthermore, scRNA-seq analysis demonstrated differential expression of these nine target genes primarily in T cells and NK cells, with CST7 showing a significant positive correlation specifically with NK cells. Beyond that, transcriptome sequencing was conducted on lung tissue collected from ALI mice, confirming the substantial differential expression of FCMR, HK3, UBE2J1, and BATF. This study provides unprecedented evidence linking the pathophysiological mechanisms of ALI/ARDS and RA to immune regulation, which offers novel understanding for future clinical treatment and experimental research.