TABLE 1.

Basic information of datasets included in this study.

Dataset ID	Number of samples	Brief introduction about the dataset	Number of deaths	Number of relapses
Discovery set	1042
GSE19188	40	A genome-wide gene expression analysis on early-stage NSCLC	24	–
GSE30219	85	Identification of a group of metastatic-prone tumors in lung cancer according to “Off-context” gene expression defined by the authors	45	27 (83)
GSE31210	226	Gene expression analysis on pathological stage I–II lung adenocarcinomas	35	64 (226)
GSE31546	16	Development of an EGFR mutation gene expression signature to predict response and clinical outcome, and identification of genes associated with the EGFR-dependent phenotype	2	–
GSE37745	106	Biomarker discovery in NSCLC	77	–
GSE50081	127	Validation of a histology-independent prognostic gene signature for early-stage NSCLC, including stage IA patients	51	37 (124)
GSE68465	442	Gene expression-based survival prediction in LUAD	236	178 (178)
Validation set	535
TCGA-LUAD	535	The LUAD cohort of TCGA, a landmark cancer genomics program, molecularly characterized over 20,000 primary cancer and matched normal samples spanning 33 cancer types.	187	–