Prediction of protein-protein interactions from amino acid sequences using a novel multi-scale continuous and discontinuous feature set

Table 1 Comparison of the prediction performance by the proposed method and some state-of-the-art works on the yeast dataset.

Model	Test set	SN(%)	PE(%)	ACC(%)	MCC(%)	AUC(%)
Our method	1	91.66	92.40	92.00	85.28	97.15
	2	90.49	92.01	91.11	83.79	96.91
	3	91.13	91.05	91.19	83.94	96.97
	4	90.41	91.48	91.01	83.64	97.07
	5	89.64	92.76	91.47	84.38	97.23
	Average	90.67 ± 0.69	91.94 ± 0.62	91.36 ± 0.36	84.21 ± 0.59	97.07 ± 0.12
Guos' work	ACC	89.93 ± 3.68	88.87 ± 6.16	89.33 ± 2.67	N/A	N/A
	AC	87.30 ± 4.68	87.82 ± 4.33	87.36 ± 1.38	N/A	N/A
Zhous' work	SVM+LD	87.37 ± 0.22	89.50 ± 0.60	88.56 ± 0.33	77.15 ± 0.68	95.07 ± 0.39
Yangs' work	Cod1	75.81 ± 1.20	74.75 ± 1.23	75.08 ± 1.13	N/A	N/A
	Cod2	76.77 ± 0.69	82.17 ± 1.35	80.04 ± 1.06	N/A	N/A
	Cod3	78.14 ± 0.90	81.86 ± 0.99	80.41 ± 0.47	N/A	N/A
	Cod4	81.03 ± 1.74	90.24 ± 1.34	86.15 ± 1.17	N/A	N/A

ISSN: 1471-2105