Machine learning approach to predict protein phosphorylation sites by incorporating evolutionary information

BMC Bioinformatics

Table 2 Three fold cross validation performance of the prediction system using ratio 1:2.

Residue	W	Ac(%)	Sn(%)	Sp(%)	Mcc	FPR
	7	78.47	29.57	95.94	0.36	0.04
	9	78.61	30.82	95.67	0.37	0.04
S	11	78.66	30.23	95.95	0.37	0.04
	13	78.78	31.53	95.66	0.38	0.04
	15	78.86	31.69	95.70	0.38	0.04
	7	74.92	34.02	94.84	0.38	0.05
	9	75.31	34.34	95.26	0.40	0.05
T	11	75.25	33.39	95.63	0.39	0.04
	13	75.30	32.76	96.01	0.40	0.04
	15	74.91	30.94	96.32	0.39	0.04
	7	71.90	12.13	99.50	0.27	0.01
	9	73.49	19.46	98.43	0.32	0.02
Y	11	72.94	17.11	98.71	0.31	0.01
	13	72.78	16.02	98.99	0.30	0.01
	15	72.40	14.21	99.27	0.29	0.01

The sensitivity (Sn) and the specificity (Sp) columns of the table reveal that the system using this ratio identifies most of the sites as negative.

ISSN: 1471-2105