A graph kernel method for DNA-binding site prediction

BMC Systems Biology

Table 3 Comparison with other methods.

Proteins	Graph Kernel¹		DISPLAR²		MV³
	Coverage	Accuracy	Coverage	Accuracy	Coverage	Accuracy
1a2p	0.45 ⁴	0.90	0.44	0.33	0.45	0.45
1ikn	0.46	0.60	0.46	0.38	0.46	0.39
1lqc	0.95	0.56	0.95	0.68	0.95	0.49
1qqi	0.70	0.67	0.75	0.66	0.70	0.33
1zzk	0.40	0.53	0.37	0.47	0.40	0.53
1qc9	0.55	0.41	0.58	0.66	0.55	0.23
2alc	0.90	0.67	0.90	0.59	0.75	0.47
1ko9	0.50	1.00	0.48	0.80	0.50	0.89
1qzq	0.57	0.63	0.57	0.24	0.57	1.00
1l3k	0.44	0.56	0.44	0.68	0.44	0.63
1xx8	0.74	0.93	0.60	0.92	0.74	0.32
1g6n	0.48	0.73	0.77	0.75	0.48	0.52
1 mml	0.46	0.35	0.38	0.50	0.46	0.40

¹ The method proposed in this study; ² The DISPLAR method developed by Tjong and Zhou [7]; ³ The MV method developed by [9]; ⁴ The bold font shows the best performance among the three methods on each protein.

ISSN: 1752-0509