Figure 1.

Flowchart for identification of a gene expression signature. Whole processes are described in five steps including pre-processing, feature selection, gene count selection, training, and validation of the prediction model. For internal validation using GSE2034, a set of gene signatures was selected and a prediction model was developed. For external validation using independent datasets (TRANSBIG), the identical gene set selected using GSE2034 was used for the features of the model. However, the parameters of the model are optimized using training data of TRANSBIG and the developed model is validated the validation data of TRANSBIG.