A two-layered machine learning method to identify protein O-GlcNAcylation sites with O-GlcNAc transferase substrate motifs

Table 3 The comparison of independent testing results between our methods and other three O-GlcNAcylation prediction tools.

Methods	TP	FN	TN	FP	Sn	Sp	Acc	MCC
Single HMM with all data	609	347	40072	20904	63.70%	65.72%	65.69%	0.076
MDD-clustered HMMs (7 OGT HMMs)	833	123	45212	15764	87.13%	74.15%	74.38%	0.171
Two-layered model (7 HMMs + 1 SVM)	828	128	51224	9752	86.61%	84.01%	84.05%	0.231
YinOYang	449	507	50619	10357	46.97%	83.01%	82.46%	0.097
O-GlcNAcScan	411	545	51219	9757	42.99%	84.00%	83.37%	0.089
O-GlcNAcPRED	554	402	38414	22562	57.95%	63.00%	62.92%	0.053

ISSN: 1471-2105