t-SNE Analysis Research Articles (Page 2)

It is well established that acute and chronic stress contributes to the onset and progression of depression, but the underlying mechanisms have not been elucidated. Here an integrated N-glycoproteomic and proteomic analysis was performed to investigate heterogeneities of glycoprotein and site-specific glycosylation between the hippocampi of control, acute stress-affected (AS), and chronic mild stress-affected (CMS) mice. 1063 unique intact N-glycopeptides, 116 N-glycan compositions, and 512 glycosylation sites were identified. CMS and AS had significant effects on glycosylation. CMS reduced multiantenna glycosylation (N8H8 and N6H5F1S1) more strongly, while AS reduced multiantenna glycosylation (N5H3F1) more strongly. CMS inhibited high-mannose synthesis with high polymerization (N2H9 and N2H8), while AS inhibited high-mannose synthesis with low polymerization (N2H6, H2H5). Furthermore, 26 and 39 glycosylation-related genes (GRGs) were identified in the AS and CMS groups, separately. Functional enrichment analysis for GRGs in the AS and CMS groups exhibited that the up-regulated functions were leading edge membrane and cell adhesion molecule binding; meanwhile, the down-regulated functions were cAMP signaling pathways. Finally, tSNE analysis based on ScRNA-seq revealed that core GRGs were highly expressed in astrocytes. All of these findings improve our understanding of glycosylation in stress-related depression, providing valuable data resources for depression pathogenesis exploration and novel therapeutic target discovery.

To propose a multimodal functional brain network (FBN) and structural brain network (SBN) topological feature fusion technique based on resting-state functional magnetic resonance imaging (rs-fMRI), diffusion tensor imaging (DTI), 3D-T1-weighted imaging (3D-T1WI), and demographic characteristics to diagnose mild cognitive impairment (MCI) in patients with unilateral middle cerebral artery (MCA) steno-occlusive disease. The performances of different algorithms on the MCI dataset were evaluated using 5-fold cross-validation. The diagnostic results of the multimodal performance were evaluated using t-distributed stochastic neighbor embedding (t-SNE) analysis. The four-modal analysis method proposed in this study was applied to identify brain regions and connections associated with MCI, thus confirming its validity. Based on the fusion of the topological features of the multimodal FBN and SBN, the accuracy for the diagnosis of MCI in patients with unilateral MCA steno-occlusive disease reached 90.00%. The accuracy, recall, sensitivity, and F1-score were higher than those of the other methods, as was the diagnostic efficacy (AUC = 0.9149). The multimodal FBN and SBN topological feature fusion technique, which incorporates rs-fMRI, DTI, 3D-T1WI, and demographic characteristics, obtains the most discriminative features of MCI in patients with unilateral MCA steno-occlusive disease and can effectively identify disease-related brain areas and connections. Efficient automated diagnosis facilitates the early and accurate detection of MCI and timely intervention and treatment to delay or prevent disease progression.

859 Background: Patients with localized unresectable or cisplatin-ineligible urothelial cancer (UC) have limited treatment options. Biomarker identification can guide targeted therapies. In the DUART study, pre-treatment immune cell subsets were significantly linked to disease control. Our planned correlative aim was to evaluate the same biomarkers using post-adjuvant treatment (post-Rx) peripheral blood mononuclear cells (PBMCs). Methods: This was a prospective, multi-institutional study BTCRC-GU15-023. Our N=16 all had valid Post-Rx values and disease control status. Eligibility criteria: >18yrs, advanced/unresectable UC, and available tumor specimen. All received concurrent durvalumab and radiation therapy followed by adjuvant durvalumab in a Phase II study. Blood samples were taken at pretreatment, 12 weeks, and post-Rx. Biomarkers were detected using multicolor flow cytometry-based analysis of PBMCs to detect T lymphocyte subsets and by dimensionality reduction using FlowJo. Correlative objective: Evaluate post-Rx time points for biomarkers that contribute to disease progression or response. Two-sample T-tests were used to study the association. All tests were two-sided and the statistical significance level used was 0.05. Results: Standard flow cytometry analysis revealed a statistically significant increase in ICOS+ CD4 and CD8 T cells in post-Rx samples among patients with progression-free survival at one year. In addition, responder patients (CR/PR/SD, n=12) showed a significant decrease in CD8 central memory T cells compared to progressors (PD, n=4) in post-Rx samples. Although not statistically significant, additional trends were noted, including decreased PD-1+ CD4 T cells in responder patients, decreased CD4 T effector memory RA+ (TEMRA), and increased CD4 naïve T cells in responder patients. There was a slight increase in interferon gamma-producing CD8 T cell subsets in responder patients and a significant decrease in central memory CD8 T cells. tSNE analysis revealed similar trends in the data, including increased naïve CD4 T cells in responder patients and slight increases in some cytokine-producing CD8 T cell subsets. Conclusions: Our small cohort demonstrates some significant differences in post-Rx T cell populations linked to therapy response, and further evaluation in a larger cohort of patients is needed. The identification of predictive biomarkers could help a more personalized therapeutic approach.

Drug non-responsiveness is the major reason for the poor prognosis of hormonal receptor-positive breast cancer (ER+/PR+ BCa), particularly the luminal A subtype. However, the underlying mechanism of drug non-responsiveness remains unknown. Flow cytometry and t-SNE analysis followed by ELISA validation of responder and non-responder unveiled lower secretion of IFN-γ, IL-12, and higher levels of IL-6 and TGF-β in CD4+ Tcells (P < 0.001), CD8+ Tcells (P < 0.001), FOXP3+ Tregs (P < 0.001) and CD206+ TAMs (P < 0.001) in non-responders. Treatment of isolated CD206+ TAMs with recombinant IL-6 upregulated the expression of ARG-1 (arginase-1) and subsequent increase of TGF-β+ Tregs (P < 0.001) and IL-6+ Tregs (P < 0.001) in luminal A BCa. Our findings showed IL-6 mediated ARG-1+CD206+ TAMs polarization induced FOXP3+ Tregs infiltration in TME of non-responder in luminal A BCa.

COPD is a healthcare problem. However, the underlying mechanism remains unclear. Our study aimed to explore the key genes involved in immune infiltration in COPD using bioinformatic tools. In this study, scRNA-seq analysis was utilized to explore specific marker genes of each immune cell subtype in COPD. TSNE analysis was used to evaluate the relationship between each immune cell cluster. Lasso regression identified 21 genes as characteristics of COPD modulated by the single-cell NK cell subpopulation. The "limma" package was used for differentially expressed analysis. The pseudotime analysis reveals the continuous changes of NK cells along their developmental trajectory. Further, we constructed a hub gene network to examine the correlation between hub genes and immune factors, transcriptional regulation factors, and potential therapeutic drugs. GO and KEGG enrichment analysis revealed the biological functions of the hub genes. RT-qPCR was used for validation of the five hub in COPD patients. NK cell subtypes are closely related to other immune cell subtypes and considered as the most important immune cells in the immune microenvironment of COPD patients. LASSO regression identified 21 genes as NK cells-characteristic genes for COPD. The GSE57148 as the training set has a AUC of 0.9489 and GSE8581 as the validation set has a AUC of 0.7303. The GO semantic similarity further confirmed five NK cell-related hub genes, C1orf56, S100A6, IGFBP7, ANXA1, and PTPN7. RT-qPCR experiment revealed that the mRNA expression of five hub genes in the normal group was lower than that in the disease group. We also found that five hub genes correlated with immune cell infiltration. The potential therapeutic agents for COPD may be zalcitabine, PP-2, PD-98059, and TGX-221 based on the CMap database prediction. We proposed that peripheral NK cells may play a role in the pathogenesis of COPD through bioinformatic analysis. These hub genes may provide insights into mechanistic research and new targets for new therapies in patients with COPD.

To decrease the number of chronic kidney disease (CKD), early diagnosis of diabetic kidney disease is required. We performed invariant information clustering (IIC)-based clustering on glomerular images obtained from nephrectomized kidneys of patients with and without diabetes. We also used visualizing techniques (gradient-weighted class activation mapping (Grad-CAM) and generative adversarial networks (GAN)) to identify the novel and early pathological changes on light microscopy in diabetic nephropathy. Overall, 13,251 glomerular images (7,799 images from diabetes cases and 5,542 images from non-diabetes cases) obtained from 45 patients in Kanazawa Medical University were clustered into 10 clusters by IIC. Diabetic clusters that mainly contained glomerular images from diabetes cases (Clusters 0, 1, and 2) and non-diabetic clusters that mainly contained glomerular images from non-diabetes cases (Clusters 8 and 9) were distinguished in the t-distributed stochastic neighbor embedding (t-SNE) analysis. Grad-CAM demonstrated that the outer portions of glomerular capillaries in diabetic clusters had characteristic lesions. Cycle-GAN showed that compared to Bowman’s space, smaller glomerular tufts was a characteristic lesion of diabetic clusters. These findings might be the subtle and novel pathological changes on light microscopy in diabetic nephropathy.

The digital recognition and preservation of historical architectural heritage has become a critical challenge in cultural inheritance and sustainable urban development. While deep learning methods show promise in architectural classification, existing models often struggle to achieve ideal results due to the complexity and uniqueness of historical buildings, particularly the limited data availability in remote areas. Focusing on the study of Chinese historical architecture, this research proposes an innovative architectural recognition framework that integrates the Swin Transformer backbone with a custom-designed Global Channel and Spatial Attention (GCSA) mechanism, thereby substantially enhancing the model’s capability to extract architectural details and comprehend global contextual information. Through extensive experiments on a constructed historical building dataset, our model achieves an outstanding performance of over 97.8% in key metrics including accuracy, precision, recall, and F1 score (harmonic mean of the precision and recall), surpassing traditional CNN (convolutional neural network) architectures and contemporary deep learning models. To gain deeper insights into the model’s decision-making process, we employed comprehensive interpretability methods including t-SNE (t-distributed Stochastic Neighbor Embedding), Grad-CAM (gradient-weighted class activation mapping), and multi-layer feature map analysis, revealing the model’s systematic feature extraction process from structural elements to material textures. This study offers substantial technical support for the digital modeling and recognition of architectural heritage in historical buildings, establishing a foundation for heritage damage assessment. It contributes to the formulation of precise restoration strategies and provides a scientific basis for governments and cultural heritage institutions to develop region-specific policies for conservation efforts.

Traditional image classification often misclassifies unknown samples as known classes during testing, degrading recognition accuracy. Open-set image recognition can simultaneously detect known classes (KCs) and unknown classes (UCs) but still struggles to improve recognition performance caused by open space risk. Therefore, we introduce a cosine distance loss function (CDLoss), which exploits the orthogonality of one-hot encoding vectors to align known samples with their corresponding one-hot encoder directions. This reduces the overlap between the feature spaces of KCs and UCs, mitigating open space risk. CDLoss was incorporated into both Softmax-based and prototype-learning-based frameworks to evaluate its effectiveness. Experimental results show that CDLoss improves AUROC, OSCR, and accuracy across both frameworks and different datasets. Furthermore, various weight combinations of the ARPL and CDLoss were explored, revealing optimal performance with a 1:2 ratio. T-SNE analysis confirms that CDLoss reduces the overlap between the feature spaces of KCs and UCs. These results demonstrate that CDLoss helps mitigate open space risk, enhancing recognition performance in open-set image classification tasks.

Objective:This study conducted an unsupervised learning cluster analysis on urine cytological images of high-grade urothelial carcinoma to assess their explanatory potential.Materials and Methods:A total of 124 urine cytology specimens of urothelial carcinoma, collected between December 2010 to December 2021 at Gunma University Hospital, were analyzed. Ten cytological image fields per specimen were captured, and pathological T factors were examined using principal component analysis and t-distributed stochastic neighbor embedding (t-SNE) with machine learning (ML) software. Common image features were also verbalized and manually reevaluated.Results:In the t-SNE analysis, the T1-dominant region was characterized by “few cells in the background,” whereas the T2-dominant region showed “many cells in the image,” “numerous neutrophils in the image,” and “abundant tumor cells in the image.” Human reassessment identified significant differences related to muscle invasion status for all findings except “abundant tumor cells in the image.” Furthermore, we confirmed that histological neutrophil infiltration was related to the abundance of neutrophils in the cytological specimens.Conclusion:This study is noteworthy as the cluster analysis identified previously unreported variations in background cell types and quality linked to muscle invasion status, and it also demonstrated the explainability of ML-derived findings through manual reassessment.

CNS embryonal tumors with PLAGL amplification (ET, PLAGL) are a recently described tumor type marked by amplification of one of the PLAG family genes, PLAGL1 or PLAGL2. Separately, a supratentorial, ependymoma-like CNS tumor type with PLAG family alteration, namely PLAGL1 fusion, was also reported (NET_PLAGL1). Here, we use DNA methylation profiling in combination with copy number, RNA-seq, and histological analysis to characterize and classify a novel group of CNS embryonal tumors harboring PLAG1 gene fusions (n=12). Through our screening, we identified a subset of CNS tumors (n=12) epigenetically distinct from other known CNS tumor types, but clustering close to the PLAGL1- and PLAGL2-amplified ET, PLAGL subtypes in our t-SNE analysis. Copy number profiles indicated putative PLAG1 fusions, which were confirmed in 9/12 tumors (not determined in 3/12). Different 5’ fusion partners (ASAP1, ADGRG1, TMEM68, TCF4, CHD7, NCALD, HNRNPK, LOC105378102) were identified that upregulate wild-type PLAG1 through promoter hijacking. Expression analysis shows upregulation of PLAG1 as well as IGF2, DLK1, Desmin, CYP2W1, and RET, which are also robustly expressed in PLAGL1/2-amplified tumors. Patient characteristics, survival data, and clinical/imaging analysis show additional similarities to PLAGL1/2-amplified tumors. Median age at diagnosis was 5 years, tumors were located throughout the neuroaxis, and original histological diagnoses were heterogeneous. The tumors demonstrated morphologic heterogeneity, with most composed of densely cellular areas of primitive small blue cells, alongside focal regions showing clear cell morphology, microcystic changes, and ependymoma-like perivascular pseudorosettes. Applied treatment regimens were also heterogeneous, but some favorable responses to therapy were observed. In summary, we describe a third subtype of PLAG family-altered pediatric CNS embryonal tumor characterized by PLAG1 gene fusion, which leads to upregulation of PLAG1 and downstream genes. We therefore propose to rename ET, PLAGL to ET, PLAG (CNS embryonal tumor with PLAG family gene alteration) together with a specification of the respective subtype.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00401-025-02917-z.

Wearable biosensors have revolutionized human performance monitoring by enabling real-time assessment of physiological and biomechanical parameters. However, existing solutions lack the ability to simultaneously capture breath-force coordination and muscle activation symmetry in a seamless and non-invasive manner, limiting their applicability in strength training and rehabilitation. This work presents a wearable smart sportswear system that integrates screen-printed graphene-based strain sensors with compact electronics for wireless data transfer and a deep learning framework for real-time classification of exercise execution quality. By leveraging 1D ResNet-18 for feature extraction, the system achieves 92.1% classification accuracy across six exercise conditions, distinguishing between breathing irregularities and asymmetric muscle exertion. Additionally, tSNE analysis and Grad-CAM-based explainability visualization confirm that the network accurately captures biomechanically relevant features, ensuring robust interpretability. The proposed system establishes a foundation for next-generation AI-powered sportswear, with applications in fitness optimization, injury prevention, and adaptive rehabilitation training.

The growing demand for advanced analytical techniques to explore complex cellular targets of nanotherapeutics has driven the development of innovative methodologies. This protocol presents a refined approach for fluorescent labeling and flow cytometric analysis of colonic cells following oral lipid nanoparticle (LNP) treatment, focusing on LNP uptake in colonic cell subpopulations in a DSS-induced colitis mouse model. By integrating optimized fluorochrome selection and gating strategies with advanced t-distributed stochastic neighbor embedding (t-SNE) analysis, this method enables precise identification and multidimensional visualization of LNP-targeted epithelial and macrophage populations under the complex conditions of inflamed colon tissue. Building on our previous studies demonstrating the effectiveness of nanoparticles in targeted drug delivery, this approach highlights the utility of flow cytometry for assessing uptake efficiency and cellular targeting. Unlike conventional protocols, it incorporates t-SNE for enhanced multidimensional analysis, allowing for the detection of subtle cellular patterns and the delineation of intricate clusters. By addressing gaps in traditional methodologies, this protocol provides a robust and reproducible framework for investigating in vivo cellular targets and optimizing drug delivery strategies for nanomedicines. Key features • This protocol is optimized for investigating nanoparticle uptake in inflamed colonic tissues from DSS-induced colitis models. • This protocol integrates flow cytometry with t-SNE for high-dimensional data analysis, enabling detailed characterization of cellular populations.

This research addresses the challenges of diagnosing and treating Autism Spectrum Disorder (ASD) using dimensionality reduction techniques and machine learning approaches. Challenges in social interaction, communication, and repetitive behaviours characterize ASD. The dimension reduction used in this research aims to identify what features influence autism cases. Several dimension data reduction techniques used in this research include PCA, Isomap, t-SNE, LLE, and factor analysis, using metrics such as Purity, silhouette score, and the Fowlkes-Mallows index. The machine learning approach applied in this study is k-medoid. By employing this method, our goal is to pinpoint the unique characteristics of autism that may facilitate the detection and diagnosis process. The data used in this research is a dataset collected for autism screening in adults. This dataset contains 20 features: ten behavioural features (AQ-10-Adult) and ten individual characteristics. The results indicate that Factor Analysis outperforms other methods based on purity metrics. However, due to data structure issues, the t-SNE method cannot be evaluated using purity metrics. PCA and LLE consistently provide stable silhouette scores across different values. The Fowlkes-Mallows index results closely align, but t-SNE tends to yield lower values. The choice of algorithm requires careful consideration of preferred metrics and data characteristics. Factor analysis is adequate for Purity, while PCA and LLE consistently perform well. This research aims to improve the accuracy of ASD identification, thereby enhancing diagnostic and treatment precision.

Patients with epidermal growth factor receptor mutant nonsmall cell lung cancer (NSCLC) often fail to treat gefitinib because of secondary drug resistance. The development of tumor drug resistance is closely related to variations in cancer cell metabolism. Single-cell metabolomics analysis can provide unique information about tumor drug resistance. Herein, we constructed a platform to study the secondary resistance of tumor cells based on single-cell metabolomics (sSRTC-scM). A gefitinib-resistant NSCLC cell line (PC9GR) was constructed by increasing the dose step by step. The metabolic profiles of parental PC9 cells and PC9GR cells with different drug resistance levels were detected by intact living-cell electrolaunching ionization mass spectrometry at the single-cell level. The data were analyzed by statistical methods such as t-SNE, variance, volcano plot, heat map, and metabolic pathway analysis. Using this platform, we found that the metabolic fingerprints of PC9GR cells can evaluate drug resistance degrees. The metabolic fingerprints continue to be altered with the increase of drug resistance. We revealed 19 metabolic markers of secondary resistance by variance analysis and clarified that the glycerophospholipid metabolic pathway of PC9GR cells changed significantly. In addition, we found that with the increase in drug resistance levels, the heterogeneity of single-cell metabolism became greater and the number of cells with weak drug resistance gradually decreased. This phenomenon can be utilized to illustrate the drug resistance degrees of PC9GR cells. This study provides diagnostic markers for evaluating the drug resistance of tumors and gives new insight into overcoming the secondary resistance of tumors.

Electric shock protection is critical for ensuring power safety in low-voltage grids, and robust fault diagnosis methods provide an essential foundation for the accurate operation of such protection devices. However, current low-voltage electric shock protection devices often suffer from limitations in operational precision and in their ability to effectively recognize electric shock types. To address these challenges, this paper proposes a fault diagnosis method for low-voltage electric shocks based on an attention-enhanced parallel CNN-BiLSTM model. The method first utilizes CNN to extract local spatial features of the electric shock signal and BiLSTM to capture temporal features. An attention mechanism is then introduced to fuse the local spatial and temporal features with weighted emphasis. Finally, a fully connected layer maps the fused features to the output layer, generating diagnostic results. Visualization through T-SNE analysis validates the improvement in model performance due to the attention mechanism. Comparative experiments show that the proposed model outperforms single models and other combined models in terms of accuracy, precision, recall, F1 score, and convergence speed. The results demonstrate that the proposed model achieves a fault diagnosis accuracy of 99.55%.

Canopy clustering is an effective method for determining the number of clusters dynamically without requiring a predefined cluster count, making it particularly suitable for large and complex datasets. However, its performance is highly dependent on the manual tuning of threshold parameters T1 and T2, which can be time-consuming and inefficient. This study aims to enhance the Canopy clustering algorithm by automating the optimization of threshold ranges using intelligent optimization algorithms. We propose a novel framework that integrates Simulated Annealing (SA), Particle Swarm Optimization (PSO), and Snake Optimization (SO) to automatically determine the optimal values of T1 and T2. Additionally, to address high-dimensional data complexity, we employ dimensionality reduction techniques such as t-SNE, SNE, and Kernel Principal Component Analysis (KPCA). The silhouette coefficient is utilized as the fitness function to evaluate clustering performance. Comprehensive experiments conducted on the Wine, Iris, and MNIST Subset datasets demonstrate that the proposed optimization-based Canopy clustering framework significantly improves clustering accuracy by up to 21% on the Wine dataset and 19% on the Iris dataset compared to traditional methods. Specifically, on the Wine dataset, the optimized Canopy clustering achieved a silhouette coefficient of 0.63, a 21% improvement over the original 0.52. On the Iris dataset, the optimized method outperformed k-means and manual Canopy clustering with silhouette coefficients of 0.62 versus 0.52 and 0.55, respectively. These results highlight the effectiveness of intelligent optimization algorithms in enhancing clustering adaptability and efficiency.

The integration of social media in recruitment and selection processes has profoundly reshaped human resource management within the Information Technology (IT) sector. Platforms such as LinkedIn, Facebook, and Twitter enable organizations to efficiently identify, assess, and engage talent by leveraging advanced search algorithms, analytics, and targeted advertising. This study explores the benefits of social media, including enhanced diversity, increased access to passive candidates, and improved employer branding. It also identifies challenges such as privacy concerns, potential biases, and ethical dilemmas. The research employs methodologies like topic modeling, t-SNE analysis, and trend analysis to uncover patterns and clusters in social media recruitment practices. Findings reveal that social media reduces hiring timelines, fosters inclusivity, and optimizes recruitment strategies. However, ethical and privacy challenges necessitate robust policies, training, and a balanced approach integrating traditional and digital recruitment methods. The study concludes with recommendations for managerial practices, societal inclusivity, and future research avenues, highlighting the transformative potential of social media in IT sector recruitment. DOI: https://doi.org/10.52783/pst.1050

Sustainability today ranks as one of the most important requirements for the operation of organisations. There is a wealth of research on the subject, but few studies have addressed the requirements of sustainability management. This study fills this gap. It aims to examine the reported practices of SME managers in three nations to assess their preparedness as regards sustainability leadership expectations. In-depth interviews with Hungarian (36), Polish (32) and Slovakian (30) SME managers were conducted using the Voyant Tools method. Among the web-based statistical analysis and visualisation software tools, Document Terms, Word Clouds, WordTree, Trends, Context, StreamGraph, ScatterPlot (t-SNE analysis, Correspondence Analysis), Correlations, and Significance were used. Ethical style, long-term planning, and elements of culture (trust, knowledge sharing, teamwork) that best qualify sustainability leadership were examined. The results show that SME managers still conduct their leadership in the spirit of traditional thinking, in a democratic style according to their assessment. Although they think about the long term, they lack vision. Building trust, a prerequisite for knowledge sharing and teamwork, appears to be difficult. Overall, the requirements of sustainable leadership are applied at different levels but are not yet common practice in any of the nations examined.

The impact of methylome analysis on the diagnosis and treatment of CNS tumours in children and adolescents: a population-based study in Greece.

BACKGROUNDGenome-wide DNA methylation analysis is currently available and is used to diagnose brain tumors. Although MGMT promoter hypermethylation is a strong biomarker in glioblastoma (GBM), there are few reports of DNA methylation profiles associated with prognosis or other DNA methylation biomarkers. In this study, we examined DNA methylation profiles correlated with prognosis and DNA methylation biomarkers. METHODSWe selected 20 long-term survivors (LS) and 20 short-term survivors (SS) from 407 patients with newly diagnosed IDH-wild-type GBM treated at our institution, divided by MGMT methylation status (methylated; M/unmethylated; U), and adjusted for age. DNA methylation analysis was performed by EPIC array, and DKFZ classifier, tSNE analysis, and copy number analysis were performed. DNA methylation data and DNA methylation profiles were compared between LS and SS groups. RESULTSThere were no significant differences in age and sex between the LS and SS groups in the M and U groups, respectively. Median OS (months) in the M group was 62.1/17.3 in LS/SS and 28.1/6.8 in the U group. Differentially methylated genes in the promoter region between LS/SS groups in the M group were HOXD13 and HAGH (methylated in LS), KCTD14 and ALPK3 (methylated in SS). Those in the U group were KLF14 and HOXB4 (methylated in LS), ZMYND10 and SLFN13 (methylated in SS). Gene Ontology analysis revealed that the genes with a hypermethylated promoter in the LS are enriched in G protein-coupled receptor signaling pathway and plasma membrane in the M group and Homeobox antennapedia and sequence-specific double-stranded DNA binding in the U group. CONCLUSIONWe identified DNA methylation biomarkers and key pathways involved in the prognosis of GBM with methylated and unmethylated MGMT promoter. Multi-omics analysis is necessary to validate the mechanism related to the prognosis of GBM.