Skip to main content

Table 2 The performance of SVM-based models developed using various features; models were evaluated on independent dataset (external cross-validation)

From: Computer-aided prediction of antigen presenting cell modulators for designing peptide-based vaccine adjuvants

Features Threshold Sensitivity (%) Specificity (%) Accuracy (%) MCC AUROC Parameters
AAC* − 0.1 94.10 ± 2.70 93.77 ± 3.28 93.91 ± 2.00 0.88 ± 0.03 0.98 ± 0.00 g = 0.001, c = 3, j = 3
N5 AAC 0 89.75 ± 3.61 90.88 ± 3.67 90.32 ± 2.27 0.81 ± 0.04 0.95 ± 0.01 g = 0.0005, c = 2, j = 1
C5 AAC 0 91.12 ± 3.53 91.64 ± 2.90 91.40 ± 2.13 0.83 ± 0.04 0.97 ± 0.01 g = 0.001, c = 9, j = 1
N5C5 AAC − 0.2 94.61 ± 3.35 93.59 ± 3.23 94.07 ± 2.16 0.88 ± 0.04 0.98 ± 0.00 g = 0.0005, c = 1, j = 1
DPC 0 93.77 ± 2.76 95.32 ± 1.40 94.64 ± 1.24 0.89 ± 0.02 0.99 ± 0.00 g = 0.0005, c = 1, j = 2
N5 DPC − 0.1 81.68 ± 4.25 87.36 ± 2.78 84.62 ± 2.65 0.69 ± 0.05 0.93 ± 0.01 g = 1e−05, c = 9, j = 1
C5 DPC − 0.1 92.31 ± 3.36 94.71 ± 2.45 93.55 ± 1.40 0.87 ± 0.02 0.98 ± 0.01 g = 0.0005, c = 1, j = 2
N5C5 DPC − 0.2 94.10 ± 3.06 93.75 ± 2.33 93.90 ± 1.49 0.88 ± 0.03 0.98 ± 0.01 g = 0.0001, c = 1, j = 1
N5 bin − 0.1 88.46 ± 2.90 89.43 ± 3.32 88.98 ± 2.36 0.78 ± 0.04 0.95 ± 0.01 g = 0.5, c = 2, j = 1
C5 bin − 0.2 93.70 ± 3.03 87.88 ± 4.25 90.63 ± 2.43 0.82 ± 0.04 0.97 ± 0.01 g = 0.5, c = 1, j = 2
N5C5 bin 0.2 90.95 ± 3.18 91.13 ± 3.44 91.03 ± 2.77 0.82 ± 0.05 0.97 ± 0.01 g = 0.05, c = 1, j = 4
N10 bin − 0.2 89.38 ± 6.68 90.46 ± 4.67 90.01 ± 3.26 0.79 ± 0.06 0.95 ± 0.03 g = 0.1, c = 2, j = 2
C10 bin − 0.2 85.02 ± 8.02 85.24 ± 5.15 85.19 ± 4.09 0.69 ± 0.09 0.93 ± 0.03 g = 0.05, c = 3, j = 1
N10C10 bin − 0.4 88.73 ± 5.95 92.33 ± 5.69 91.04 ± 2.52 0.81 ± 0.05 0.97 ± 0.02 g = 0.1, c = 1, j = 1
AAC + motif − 0.1 93.11 ± 1.86 95.33 ± 3.13 94.35 ± 1.67 0.89 ± 0.03 0.99 ± 0.00 g = 0.001, c = 6, j = 1
DPC + motif 0 93.28 ± 2.38 96.36 ± 1.70 95.00 ± 1.25 0.90 ± 0.02 0.99 ± 0.00 g = 0.0005, c = 1, j = 2
  1. This table shows average performance (mean ± standard deviation) of models on randomly generated independent datasets (bagging)
  2. MCC Matthews correlation coefficient, AAC amino acid composition, DPC dipeptide composition, N5 first 5 residues from N terminus, C5 first 5 residues from C terminus, N5C5 first 5 residues from N and C terminus respectively, bin binary profile, AAC + motif amino acid composition with MERCI motif score, DPC + motif dipeptide composition with MERCI motif score, SVM parameters g gamma parameter of the radial basis function, , c trade-off between training error and margin, j regularization parameter (cost-factor, by which training errors on positive examples outweigh errors on negative examples)