BMC Bioinformatics

Table 1 Accuracy and coverage of the present approach for predicting kinase - substrate interactions

From: Modular composition predicts kinase/substrate interactions

		Training				Testing
		PTM	2006	HPRD		BioGrid		PTM	2009
		TK	S/TK	TK	S/TK	TK	S/TK	TK	S/TK
Data	EPPI	886	2925	137	199	111	166	33	237
	Kinase	56	176	60	104	41	67	15	61
	Substrate	274	881	32	44	26	44	21	90
	VPPI	15344	155056	1920	4576	1066	2948	315	5490
Present	PPPI	1132	7133	43	204	36	69	27	193
	MPPI	617	1876	18	26	17	15	6	16
	p value	0	0	9.0 E-12	1.3E-07	1.0E-09	7.0E-07	0.014	0.0038
	Se	69.6	64.1	13.14	13.07	15.32	0.04	18.18	6.75
	Sp	96.6	95.4	97.76	95.54	96.62	97.66	91.43	96.48
CSSM	PPPI	14496	122975	763	1411	433	836	234	4462
	MPPI	826	2815	84	67	64	53	26	160
	p value	0.9416	0	3.79E-08	0.0096	4.33E-05	0.128	0.204	1
	Se	93.23	96.24	61.31	33.67	57.66	31.93	78.79	67.51
	Sp	5.44	20.69	60.26	69.17	59.38	71.64	25.71	18.72
K-GIDDI	PPPI	1491	1557	117	75	88	68	33	65
	MPPI	206	42	19	6	17	5	2	11
	p value	0	0.0098	1.32E-04	0.0432	0.0025	0.1814	0.7	0.0002
	Se	23.25	1.44	13.87	3.02	15.32	3.01	6.06	4.64
	Sp	90.28	99	93.9	98.36	91.74	97.69	89.52	98.81

EPPI: Number of Experimental PPI; PPPI: Number of predicted PPI; MPPI: Number of matches between PPPI and EPPPI; VPPI: Number of Virtual PPI used in p value computations. Also shown is the prediction coverage and accuracy of two previously published approaches (CSSM, K-GIDDI).

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com