The effect of three novel feature extraction methods on the prediction of the subcellular localization of multi-site virus proteins

Figures & data

Table 1. The benchmark dataset S taken from Virus-mPloc²¹.

Subset	Subcellular location	Number of proteins
S1	Host viral capsid	8
S2	Host cell membrane	33
S3	Host endoplasmic reticulum	20
S4	Host cytoplasm	87
S5	Host nucleus	84
S6	Secreted	20
Total number of locative proteins		252
Total number of different proteins		207

Figure 1. (a) The impact of each residue on the subsequent residues. Fig. 1(b). The impact of each residue on the subsequent residues. Fig. 1(c). The impact of each residue on the subsequent residues.

Table 2. Sorting signals of database.

(0,30)	(1,30)	(2,30)	(3,30)	(4,30)	(5,30)	(6,30)	(7,30)
50%	45.63%	50%	54.76%	57.54%	53.57%	57.54%	47.22%
(8,30)	(9,30)	(10,30)	(11,30)	(12,30)	(13,30)	(14,30)	(15,30)
55.56%	46.43%	44.05%	41.27%	49.21%	50%	43.25%	50.79%
(16,30)	(17,30)	(18,30)	(19,30)	(20,30)	(21,30)	(22,30)	(23,30)
50.79%	51.98%	55.56%	56.35%	46.83%	46.03%	46.03%	44.05%

Table 3. Application of two methods to original database and new database.

AAC in original dataset	R-AAC	Dipeptide in original dataset	R-Dipeptide
55.16%	58.33%	54.76%	59.92%

Table 4. Six physicochemical properties.

hydrophobicity	hydrophilicity	Pk1	Pk2	PI	mass
0.62	−0.5	2.35	9.87	6.11	15
0.29	−1	1.71	10.78	5.02	47
−0.9	3	1.88	9.6	2.98	59
−0.74	3	2.19	9.67	3.08	73
1.19	−2.5	2.58	9.24	5.91	91
0.48	0	2.34	9.6	6.06	1
−0.4	−0.5	1.78	8.97	7.64	82
1.38	−1.8	2.32	9.76	6.04	57
−1.5	3	2.2	8.9	9.47	73
1.06	−1.8	2.36	9.6	6.04	57
0.64	−1.3	2.28	9.21	5.74	75
−0.78	0.2	2.18	9.09	10.76	58
0.12	0	1.99	10.6	6.3	42
−0.85	0.2	2.17	9.13	5.65	72
−2.53	3	2.18	9.09	10.76	101
−0.18	0.3	2.21	9.15	5.68	31
−0.05	−0.4	2.15	9.12	5.6	45
1.08	−1.5	2.29	9.74	6.02	43
0.81	−3.4	2.38	9.39	5.88	130
0.26	−2.3	2.2	9.11	5.63	107

Table 5. Application of PseAAC, R-Dipeptide, I-PseAAC and PseAAC2 to the new database.

PseAAC	R-Dipeptide	I-PseAAC	PseAAC2
57.14%	59.92%	59.13%	57.94%

Table 6. Overall accuracy of R-Dipeptide, I-PseAAC, and PseAAC2.

	Overall accuracy
Subcellular location	R-Dipeptide	I-PseAAC	PseAAC2
Viral capsid	7/8 = 87.5%	7/8 = 87.5%	7/8 = 87.5%
Host cell membrane	12/33 = 36.36%	13/33 = 39.39%	13/33 = 39.39%
Host endoplasmic reticulum	11/20 = 55%	11/20 = 55%	10/20 = 50%
Host cytoplasm	49/87 = 56.32%	52/87 = 59.77%	51/87 = 58.62%
Host nucleus	59/84 = 70.24%	51/84 = 60.71%	52/84 = 61.9%
Secreted	13/20 = 65%	15/20 = 75%	13/20 = 65%
Overall accuracy	151/252 = 59.92%	149/252 = 59.13%	146/252 = 57.94%

Table 7. Evaluation functions for PseAAC, R-Dipeptide, I-PseAAC, and PseAAC2.

FEM	A	C	H	O	R
PseAAC	0.7662	0.1798	0.1428	0.3888	0.1396
R-Dipeptide	0.7838↑	0.1712↓	0.1296↓	0.3571↓	0.1293↓
I-PseAAC	0.7684↑	0.1798	0.1355↓	0.3809↓	0.1396
PseAAC2	0.7686↑	0.1772↓	0.1329↓	0.3849↓	0.1365↓

Wang L, Wang D, Chen Y, Qiao S, Zhao Y, Cong H. Feature Combination Methods for Prediction of Subcellular Locations of Proteins with Both Single and Multiple Sites. Intelligent Computing Theories and Application. 2016;9771:192-201. doi:10.1007/978-3-319-42291-6_19.

 Google Scholar