Annotation of gene promoters by integrative data-mining of ChIP-seq Pol-II enrichment data

BMC Bioinformatics

Table 2 Performance statistics of classification models and other existing programs based on independent test set

Method	Promoters = 2980, NonPromoters = 11481
	# of true positive	# of false negative	# of true negative	# of false positive	Sensitivity (%)	Positive predictive value (PPV)%	Mathew correlation coefficient	True positive cost
Bagging	2593	387	11385	96	87.01	96.43	0.9	0.04
LogitBoost	2594	386	11356	125	87.05	95.4	0.89	0.05
Random Forest	2599	381	11349	132	87.21	95.17	0.89	0.05
Rotational Forest	2391	589	11332	149	80.23	94.13	0.84	0.06
EP3 Program	2493	487	11064	417	86.91	85.67	0.81	0.17
Eponine Program	2581	399	9633	1848	87.01	58.28	0.62	0.72
ProSOM	2563	417	8817	2664	86.01	49.03	0.53	1.04
FirstEF	1714	1226	11402	79	57.52	95.6	0.70	0.05

ISSN: 1471-2105