PSO-LocBact: A Consensus Method for Optimizing Multiple Classifier Results for Predicting the Subcellular Localization of Bacterial Proteins
Table 4
Accuracy of PSO-LocBact in different experimental cases.
Gram-negative bacterial proteins
Location:
Extracellular region (%)
Outer membrane (%)
Periplasm (%)
Inner membrane (%)
Cytoplasm (%)
Overall (%)
Experimental case 1
98.84
100
93.02
97.67
100
97.67
Experimental case 2 (>90)
100
100
87.21
100
100
97.44
Experimental case 3 (<90)
80.23
91.86
90.69
94.19
96.51
90.69
Experimental case 4 (>80)
100
100
89.53
100
98.84
97.67
Experimental case 5 (<80)
84.88
54.65
84.88
95.35
94.19
82.79
Experimental case 6 (>70)
100
100
94.19
100
100
98.84
Experimental case 7 (<70)
44.19
66.28
81.4
88.37
95.35
75.12
Experimental case 8 (>60)
100
100
94.19
100
98.83
98.60
Experimental case 9 (<60)
76.74
96.51
86.05
81.39
93.02
86.74
Gram-positive bacterial proteins
Location:
Extracellular region (%)
Cell wall (%)
Inner membrane (%)
Cytoplasm (%)
Overall (%)
Experimental case 1
96.05
100
100
98.73
98.71
Experimental case 2 (>90)
96.34
96.39
100
100
98.20
Experimental case 3 (<90)
89.02
54.22
98.82
100
85.63
Experimental case 4 (>80)
97.56
96.39
100
100
98.50
Experimental case 5 (<80)
78.05
43.37
91.76
100
78.44
Experimental case 6 (>70)
94.73
100
100
100
98.71
Experimental case 7 (<70)
68.29
12.05
100
100
70.66
Experimental case 8 (>60)
97.56
96.39
100
100
98.50
Experimental case 9 (<60)
NA
NA
NA
NA
NA
Experimental case 1: performance of PSO-LocBact without PSORTb 3.0. Experimental case 2: performance of PSO-LocBact considering only classifiers with accuracy ≥90%. Experimental case 3: performance of PSO-LocBact considering only classifiers with accuracy <90%. Experimental case 4: performance of PSO-LocBact considering only classifiers with accuracy ≥80%. Experimental case 5: performance of PSO-LocBact considering only classifiers with accuracy <80%. Experimental case 6: performance of PSO-LocBact considering only classifiers with accuracy ≥70%. Experimental case 7: performance of PSO-LocBact considering only classifiers with accuracy <70%. Experimental case 8: performance of PSO-LocBact considering only classifiers with accuracy ≥60%. Experimental case 9: performance of PSO-LocBact considering only classifiers with accuracy <60%.