Research Paper Volume 15, Issue 14 pp 6798—6821

Figure 7. The prognostic model was constructed based on DNA sensors related genes. (A) LASSO coefficient profiles of 14 prognostic mRNAs in TCGA training cohort and the coefficient profile plot was developed against the log (Lambda) sequence. When lambda is 0.0423, the smallest value, fourteen genes are obtained. (B) Multi-variate analyses. (C) Analyses of KM and ROC on TCGA training data. (D) The KM and ROC analysis of the model on the TCGA validation dataset. (E) The KM and ROC analysis of the model on the TCGA dataset. (F) The KM and ROC analysis of the model on the HCCDB18 dataset. (G) The KM and ROC analysis of the model on the GSE76427 dataset. The asterisks represent the P-value (*p < 0.05, **p < 0.01). The statistical analyses were performed by the Kaplan-Meier analysis. LASSO, Least absolute shrinkage and selection operator; TCGA, the cancer genome atlas; KM, Kaplan–Meier; ROC, receiver operating characteristic.