Table 3. Predictive performance of SPIA, GGEA and MinePath on three BrCa/ER datasets utilizing single-relations Rows and columns refer to training and test data, respectively.

Abbreviations are the same with those used in Table 2; bold figures indicate superior performance.

SPIA
Discriminant SPs	GEO dataset	GSE2034		GSE3494		GSE7390		Average ACC%
		ACC%	AUC	ACC%	AUC	ACC%	AUC
569	GSE2034	—	—	86.2	0.549	79.3	0.708	82.8
414	GSE3494	79.4	0.637	—	—	71.7	0.571	75.6
469	GSE7390	80.1	0.691	91.5	0.815	—	—	85.8
Average:	484.0
GGEA
Discriminant SPs	GEO dataset	GSE2034		GSE3494		GSE7390		Average ACC%
		ACC%	AUC	ACC%	AUC	ACC%	AUC
1006	GSE2034	—	—	87.9	0.559	83.8	0.770	85.9
693	GSE3494	79.7	0.722	—	—	84.8	0.881	82.3
525	GSE7390	77.6	0.630	89.9	0.632	—	—	83.8
Average:	741.3
MinePath
Discriminant SPs	GEO dataset	GSE2034		GSE3494		GSE7390		Average ACC%
		ACC%	AUC	ACC%	AUC	ACC%	AUC
1271	GSE2034	—	—	89.9	0.657	68.7	0.516	79.3
2375	GSE3494	75.5	0.562	—	—	71.2	0.555	73.4
2917	GSE7390	83.9	0.722	87.0	0.740	—	—	85.5
Average:	2187.7