Table 1.

Summary of dataset selection

Database	GEO	ArrayExpress			GLIOMASdb
Query	»glioblastoma«	»glioblastoma, grade«	»glioblastoma, stem cells«	»glioblastoma, neurospheres«	/
Filters	»homo sapiens«, »DataSets«	»homo sapiens«, »RNA assay«, »array assay«			/
All datasets	36	76	69	14	1
Datasets > 20 samples	11	58	17	3	1
Rejected—Reason 1	5	35	11	1	0
Rejected—Reason 2	2	16	3	2	0
Rejected—Reason 3	0	4	1	0	0
Appropriate datasets	4	3	1 + 1 conditionally *	0	1
Appropriate datasets (list)	GSE4412/GDS1975 GSE4412/GDS1976 GSE4290/GDS1962 GSE23806/GDS3885	E-GEOD-16011 E-GEOD-52009 E-GEOD-68848	E-GEOD-4536 E-GEOD-74571 *	/	GLIOMASdb

Reason 1—no relevant comparison possible (Comparisons relevant for selection of new GB/GSC marker candidates are: (1) glioblastoma multiforme tissue vs. non-malignant tumor tissue, (2) glioblastoma multiforme tissue vs. tissue from low-grade—i.e., grade II—gliomas, (3) glioblastoma multiforme tissue vs. tissue from oligodendrogliomas, (4) GSCL vs. ordinary glioblastoma multiforme tissue, (5) GSCL vs. CGCL, (6) neurospheres vs. ordinary glioblastoma multiforme tissue, (7) neurospheres vs. CGCL; the group in which elevated expression of GB/GSC markers is expected is marked in bold). Reason 2—relevant comparisons possible, but not with at least 10 samples in both compared groups. Reason 3—dataset is not original (duplicates or combinations of other GEO and ArrayExpress datasets). * two relevant comparisons possible: the first compares two groups with 9 samples each, the second compares a group of 10 samples with a group of 9 samples. Conditionally acceptable because this is the only found dataset that enables comparisons including glioblastoma neurospheres.