Figure 7. cg00044170, FOXM1 and the hybrid signature for diagnosis of ccRCC. (A) Receiver operating characteristic (ROC) curves of eQTM (cg00044170 of FOXM1) in classifying all/localized tumors and normal. (B) ROC curves of FOXM1 (C) The calculation of epsilon value to simultaneously consider expression of genes and methylation of eQTMs. (D) ROC curves of eQTM-gene hybrid signature (cg00044170 and FOXM1) in classifying all/localized tumors and normal. (E) The distribution of epsilon values of patients in the training set. The median epsilon value was used as cut-off point to divide ccRCC patients into high- and low-group. The datasets named “tumors” (the whole dataset) are stage I/II tumors and stage III/IV tumors, while the datasets named “localized tumors” (the subset) are stage I/II tumors.