Evaluating tools for transcription factor binding site prediction

BMC Bioinformatics

Table 1 Performance of TFBS prediction methods using JASPAR.2010 PWMs

	Sn	PPV	ACC _g	FPR _s
CLUSTER
Baycis	0.599	0.497	0.545	0.040
Cister	0.635	0.565	0.599	0.037
MCast	0.774	0.682	0.726	0.032
Comet	0.682	0.589	0.634	0.037
ClusterBuster	0.656	0.580	0.617	0.036
INDIVIDUAL
Matrix-Scan	0.647	0.579	0.612	0.027
Clover	0.674	0.584	0.627	0.022
FIMO	0.816	0.734	0.774	0.015
Patser	0.723	0.653	0.687	0.016
PossumSearch	0.708	0.635	0.670	0.019

Average sensitivities (Sn), Positive Predictive Value (PPV) and geometric accuracy (ACC _g) are reported together with the false positive rate using scrambled sequences (FPR _s). The best-performing tools, MCast and FIMO are highlighted in bold. Performance was evaluated using the 14 PWMs in the PAZAR-J dataset

ISSN: 1471-2105