A gene signature based method for identifying subtypes and subtype-specific drivers in cancer with an application to medulloblastoma

Table 3 The subtype signatures of the non-Wnt/non-Shh subtypes by previous studies.

Datasets and original labels	Subtype WNT (# s.g.)	Subtype SHH (# s.g.)	NWS subtypes (# s.g.)
Cho73
1	313	243	34 *
2 5 7	342	237	45 *, ‡, ¶
4	366	222	75 †, ‡, ¶
5	339	244	93 †, ¶
7	330	245	6
(Total overlaps)	(203)	(127)	(1)
(Selective overlaps) Northcott90			(* 7, † 34, ‡ 25, ¶12)
Group C	330	179	170
Group D	396	181	186
(Total overlaps)	(253)	(112)	(76)
Kool62
Subtype C	773	287	284 *,‡
Subtype D	738	299	237 *, †, ‡
Subtype E	657	268	105 †
(Total overlaps)	(498)	(167)	(46)
(Selective overlaps)			(* 158, † 66, ‡ 53)

Each row represents the numbers of signature genes for the WNT, SHH and one of the NWS subtypes (the label of which is shown on the first column of each row), respectively. The total overlaps refer to the intersection of all signatures of a group in each dataset, e.g., |S_{Subtype C} ∩ S_{Subtype D} ∩ S_{Subtype E}| = 46. Selective overlaps refer to the intersection of the signatures with the same symbol. The difference between this table and Table S1 is that in the latter, all subtypes are trained together, while in the former, only one of the reported NWS subtypes is trained at a time with the WNT, SHH and normal cases for detection of signatures. s.g.=signature genes.

ISSN: 1471-2105