MiRTif: a support vector machine-based microRNA target interaction filter

BMC Bioinformatics

Table 4 A comparison between MiRTif and three ab initio target predicting software programs, PicTar, miRBase and TargetScan using miR-224, which was discovered to be significantly up-regulated in hepatocellular carcinoma patients.

Gene	Entrez Gene ID	Gene description	PicTar ranking	miRBase ranking	TargetScan ranking	MiRTif
H3F3B	3021	H3 histone, family 3B (H3.3B)	1	na	122	+1.69
API5	8539	Apoptosis inhibitor 5	2	na	99	+1.32
ARMCX2	9823	Armadillo repeat containing, X-linked 2	4	198	na	-0.98
CDK9	1025	Cyclin-dependent kinase 9 (CDC2-related kinase)	9	25	na	-0.15
NCOA6	23054	Nuclear receptor coactivator 6	10	204	47	+0.94
ATF2	1386	Activating transcription factor 2	64	19	104	+1.27
NUP153	9972	nucleoporin 153 kDa	168	1	46	-1.46
FOSB	2354	FBJ murine osteosarcoma viral Nncogene homolog B	222	4	125	+0.11

Ranking indicates the rank predicted by respective software program. "NA" indicates that the particular program did not produce an interaction between the corresponding target gene and miR-224. For the discriminant scores under MiRTif, positive scores indicate true predictions while negative scores negative predictions. The score is proportional to the sample's distance from the hyperplane. So a large positive value implies high confidence that the sample lies in the positive class.

ISSN: 1471-2105