Predicting phosphorylation sites using machine learning by integrating the sequence, structure, and functional information of proteins

Table 6 Performance comparison of different existing tools for pS/pT/pY site prediction

Phosphorylation site	Methods	Sensitivity (%)	Specificity (%)	MCC	AUC
Serine	PhosPred-RF	79.70	75.00	0.54	0.85
	PhosphoSVM	44.43	94.04	0.29	0.84
	PPRED	32.27	91.6	0.16	0.75
	iPhos-PseEn	79.64	79.78	0.39	–
	Our RF model	89.4	88.9	0.78	0.95
Threonine	PhosPred-RF	73.80	72.60	0.46	0.81
	PhosphoSVM	37.31	94.99	0.25	0.81
	PPRED	34.32	83.65	0.09	0.65
	iPhos-PseEn	71.51	80.68	0.34	–
	Our RF model	97.8	74.4	0.77	0.97
Tyrosine	PhosPred-RF	72.70	64.00	0.36	0.76
	PhosphoSVM	41.92	87.34	0.20	0.73
	PPRED	43.04	82.65	0.16	0.70
	iPhos-PseEn	76.18	76.29	0.32	–
	Our RF model	100	63.5	0.57	0.99

ISSN: 1479-5876