Figure 5

Identification of key genes associated with diagnosis and prognosis of thyroid cancer. (A) Expression signatures of 13 genes from the TCGA-THCA dataset based on THCA-related clusters selected in the LASSO models (left). Cross-validation for tuning parameter selection in the LASSO model (right). (B) Expression signatures of 34 genes from the merged dataset based on THCA-related clusters selected in the LASSO models (left). Cross-validation for tuning parameter selection in the LASSO model (right). (C) Identification of overlapping genes from the Venn diagram. THEMIS2 expression between tumor and normal groups of the TCGA-THCA dataset (D), merged dataset (E), and single dataset (F). (G) GSEA for THEMIS2. (H) ROC curves and AUC statistics to evaluate the capacity of discrimination of tumor specimens from healthy controls showing excellent sensitivity and specificity. (I) Kaplan–Meier survival analysis for patients with thyroid cancer stratified by expression signature groups in the THCA dataset.