Table 1.

Summary of datasets and data processing.

	Manual Screening		Computational Screening			Result	T	N	M
NCBI GEO	GSE screened: 3180 datasets	Primary tissue series n = 554 (38,897 Samples)	Data cleaning	MAS5 [10] normalization and scaling	JetSet [11] Annotation	38,431 Samples 38 tumor types	29,376	3691	453
TARGET	1193 samples	-	Data cleaning	DESeq2 [12] normalization and scaling	AnnotationDBI [13] annotation	1193 samples 7 tumor types	1180	12	1
TCGA	11,050 samples	Removal of non-primary tissues	Data cleaning	DESeq2 normalization and scaling	AnnotationDBI annotation	11,010 samples 33 tumor types	9886	730	394
GTEx	11,688 samples	Removal of non-primary tissues	Data cleaning	DESeq2 normalization and scaling	biomaRt [14] andAnnotationDBI annotation	11,215 samples 51 tumor types	-	11,215	-