Protein-protein interaction prediction using a hybrid feature representation and a stacked generalization scheme

BMC Bioinformatics

Table 8 Performance results of 10-fold CV, using mixed positive and negative data from different datasets

	PPI-MetaGO							Other recent prediction tools
Dataset	TPR	FPR	Prec	ACC	F-score	MCC	AUC	TPR	FPR	Prec	ACC	F-score	MCC	AUC	Tool
HS5(+)^a HS4(−)	0.811	0.031	0.893	0.932	0.850	0.808	0.971	0.730	0.054	0.808	0.894	0.766	0.700	0.932	HVSM
SC6(+)^b SC2(−)	0.810	0.155	0.858	0.826	0.832	0.656	0.901	0.819	0.204	0.824	0.811	0.822	0.621	0.891	DeepSequencePPI

TPR true positive rate, FPR false positive rate, Prec precision, ACC accuracy, MCC Matthews correlation coefficient, AUC area under ROC
HS Homo sapiens, SC Saccharomyces cerevisiae
^aCombination of positive data from HS5 and negative data from HS4
^bCombination of positive data from SC6 and negative data from SC2

ISSN: 1471-2105