Low GBX2 expression is associated with histological type B thymoma: a multi-omics analysis

Figures

• 

  Fig. 1. Overview of the study design for identifying GBX2, a critical gene associated with the predictive histological type and immunosuppressive phenotype of thymoma

• 

  Fig. 2. MR-DEG related module construction and biological function and pathway annotation. A Clustering dendrogram formed by weighted correlation coefficients to cluster genes with similar expression patterns into co-expression modules, with each color representing a module. B Heatmap of correlation of MEs with Masaoka stage (Stage) and sample type. C, D Top biological pathways of genes in yellow modules

• 

  Fig. 3. Identification of molecular clusters associated with MR-DEGs. A, B Selection of the optimal number of clusters for MR-DEG consensus clustering in the TCGA cohort (k = 2) C PCA analysis proves the robustness of clustering results D Heat map of the distribution of GBX2 with clinicopathological features E Screening of genes associated with clusters of relative importance > 1.0 using the Random Forest algorithm F Determination of the optimal lambda when the partial likelihood of deviance reaches a minimum G The results of screening two machine models (***p < 0.001; **p < 0.01; *p < 0.05; blank, not significant)

• 

  Fig. 4. Determine key genes that predict histological groups by machine learning. A, B The area under the ROC curve of Adaboost,LR,RF,SVM,XGBoost models in the training set and test set C, D Calibration curve in in the training set and test set E Variable SHAP value ranking F SHAP summary plot for the 8 gene in XGBoost model (SHAP, Shapley Additive exPlanations. The X-axis indicates the model’s contribution, where > 0 denotes a positive contribution and < 0 signifies a negative contribution) G The waterfall plot shows the impact weights of each gene of predicting thymoma histological groups in a single sample (The X-axis indicates the probability of this sample being a Type 2 thymoma, The X-axis indicates each gene expression in this sample) H Importance ranking of 8 genes in each models I Ranking of the area of each gene in the radar chart

• 

  Fig. 5. Validation of GBX2 correlation with thymoma histological type. A Histological type composition of the TCGA cohort in different GBX2 expression levels B, C The GBX2 expression difference in histological groups and each histological types of thymoma in TCGA cohort D Histological type composition of the GEO cohort in different GBX2 expression levels E, F The GBX2 expression difference in histological groups and each histological types of thymoma in GEO cohort G Differences in GBX2 expression between thymic tissue and tumor H Protein–Protein Interaction Network of GBX2 and MR-DEG (***p < 0.001; **p < 0.01; *p < 0.05; ns not significant)

• 

  Fig. 6. GTF2I mutation and GBX2 at the DNA level. A TCGA cohort mutation landscape B Differences in mutation frequencies among groups with different GBX2 expression levels C Composition of histological types of GTF2I mutation group D, E GBX2 methylation and mRNA expression difference between groups with GTF2I SNV mutation F, G GBX2 methylation and mRNA expression difference between groups with GTF2I copy number mutation H–M Box plots showing differences in DNA damage measurements among HGE group and LEG group (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05; ns not significant)

• 

  Fig. 7. The Role of GBX2 in the tumor microenvironment A Relationship between 27 immune cells abundance and GBX2 expression. B Immunosuppression scores in the HGE and LGE group C Abundance of tumor markers between groups with different levels of GBX2 expression D–I Negative correlation between key immune checkpoint expression (PD-L1,PD-L2,CTLA4) and GBX2 expression in TCGA cohort and GEO cohort (****p < 0.001; ***p < 0.01; *p < 0.05; ns not significant)

• 

  Fig. 8. Relationship between GBX2 and C19MC microRNA19 A Regulatory relationship between GBX2 and microRNAs in clusters on chromosome 19 B Positive correlation between GBX2 expression and microRNA expression (Cor: Correlation) C Differences in microRNA expression between groups with different levels of GBX2 expression

• 

  Fig. 9. Carcinogenesis pathway function analysis of GBX2 A Top 10 carcinogenesis pathways ranked by correlation B Differences in carcinogenesis pathway scores in the HGE and LGE group (Blue: Negative correlation,Red: Positive correlation) C The expression levels of genes encoding oxidative phosphorylation pathway proteins in the KEGG database in the TCGA cohort (7.118: Cytochrome bc₁ complex; 7.119: Cytochrome-c oxidase) D The expression levels of genes encoding oxidative phosphorylation pathway proteins in the KEGG database in the GEO cohort E Composition and distribution of single cells from SCP1532 F The expression profile of GBX2 for each cell

• 

  Fig. 10. Drug sensitivity analysis of GBX2 A IC50 values of each drug at different GBX2 expression levels (****p < 0.0001; ***p < 0.001; **p < 0.01) (IC50: Half-maximal inhibitory concentration) B Molecular docking between GBX2 and Erlotinib. The figure represents surface representation structure. Erotinib showed maximum binding energy in GBX2 (− 5.02 kcal/mol) with hydrogen bond formation at MET-48 in GBX2. The dark dashed lines represent hydrogen bonds, and the interaction distances are indicated next to the bonds