A feature selection approach for identification of signature genes from SAGE data

BMC Bioinformatics

Table 2 Summary of comparisons. For each comparison, this table shows the 10 most frequent genes along the first 50 best triples ranked by bolstered error and score. Avg. Error = bolstered error; Avg. Dist. = average distance of the nearest point from the hyperplane; std. dev. = standard deviation.

Comparison	normal × glio	normal × astro, glio	astro 2 × astro 3	astro 2 × astro3, glio	astro 3 × glio	astro 2, astro3 × glio
Ranked by avg. bolstered error
Avg. Error (std. dev.)	0.0 (0.0)	0.0 (0.0)	0.003 (0.0)	0.016 (0.003)	0.022 (0.004)	0.031 (0.003)
Avg. Dist. (std. dev.)	59.217 (90.175)	21.76 (0.35)	2.194 (0.297)	1.095 (0.222)	1.115 (0.256)	0.785 (0.199)
1	ITPKA	CPNE7	SCN4B	LOC646999	COPS5	COPS5
2	CPNE7	AK055475	GPX2	RPM2B	WASF1	ZDHHC22
3	NDFIP2	TAIP-2	ZNF233	AK090819	SAP18	BC035881
4	KIAA1345	PCDH9	APBA3	RALGPS1	VPS35	WASF1
5	HLF	LCE2D	HSU79275	RP1-32F7.2	CHGB	PRPF39
6	FLJ31636	AK095013	LRRC50	HSU79275	FLJ31818	LHFPL2
7	BX648951	C1QL2	HNF4G	LRRC50	MORF4L1	ZNF644
8	CYP7B1	LOC36003	RAET1E	FLJ20323	FLJ39538	POLDIP3
9	BC042456	GPR97	AX090819	DUSP9	ZNF233	ZBTB5
10	PIK3C3	GSTO2	MRC2	WDR35	DNASE2	LRP12
Ranked by score
Avg. Error (std. dev.)	0.0 (0.0)	0.0 (0.0)	0.006 (0.003)	0.021 (0.006)	0.028 (0.007)	0.033 (0.005)
Avg. Dist (std. dev.)	18.48 (1.17)	10.03 (0.23)	1.91 (0.23)	0.91 (0.32)	1.02 (0.28)	0.83 (0.21)
1	RPH3A	CALM3	APBA3	RRM2B	WASF1	COPS5
2	ITPKA	SEPT5	PARVA	RALGPS1	SAP18	ZDHHC22
3	HLF	EEF1A2	BC023565	PDE8A	COPS5	BC035881
4	NDFIP2	LCE2D	BC015762	BDP1	ZDHHC22	SAP18
5	SH3GL2	PPP2R4	DNASE2	RP1-32F7.2	BC035881	ANAPC13
6	SYT13	SULT4A1	KBTBD6	TMEFF1	LRP4	LRP4
7	BCL2L2	ATP1A3	KCTD9	BC023565	JPH4	JPH4
8	MGC34830	VMP	ZNF354C	BC015762	CHGB	PRPF39
9	PLEKHB2	PNMA6A	C15orf29	ZNF354C	ANAPC13	LHFPL2
10	DSCR1L1	FBXO2	SHOC2	C15orf29	POLDIP3	SSR3

ISSN: 1471-2105