Skip to main content

Table 3 Five-fold cross-validation results of the SVM models trained with various features for discriminating between positive and negative training datasets

From: MDD-carb: a combinatorial model for the identification of protein carbonylation sites with substrate motifs

Residue Training features Sn Sp Acc MCC AUC
K Amino acid composition (AAC) 0.70 0.69 0.69 0.37 0.78
Amino acid pairs composition (AAPC) 0.66 0.65 0.65 0.29 0.71
Amino acid sequence (AA) 0.68 0.64 0.65 0.23 0.67
Positional weighted matrix (PWM) 0.74 0.67 0.69 0.37 0.78
Position specific scoring matrix (PSSM) 0.63 0.61 0.62 0.16 0.61
BLOSUM62 (B62) 0.63 0.60 0.61 0.15 0.59
R Amino acid composition (AAC) 0.66 0.63 0.64 0.28 0.70
Amino acid pairs composition (AAPC) 0.62 0.61 0.61 0.22 0.65
Amino acid sequence (AA) 0.62 0.62 0.62 0.17 0.62
Positional weighted matrix (PWM) 0.71 0.70 0.70 0.39 0.80
Position specific scoring matrix (PSSM) 0.61 0.56 0.58 0.14 0.59
BLOSUM62 (B62) 0.62 0.62 0.62 0.17 0.62
T Amino acid composition (AAC) 0.74 0.70 0.72 0.41 0.82
Amino acid pairs composition (AAPC) 0.69 0.68 0.69 0.35 0.75
Amino acid sequence (AA) 0.63 0.62 0.62 0.18 0.63
Positional weighted matrix (PWM) 0.69 0.67 0.68 0.32 0.73
Position specific scoring matrix (PSSM) 0.65 0.65 0.65 0.29 0.70
BLOSUM62 (B62) 0.58 0.50 0.53 0.08 0.53
P Amino acid composition (AAC) 0.72 0.70 0.70 0.39 0.80
Amino acid pairs composition (AAPC) 0.68 0.64 0.65 0.30 0.71
Amino acid sequence (AA) 0.64 0.66 0.65 0.23 0.67
Positional weighted matrix (PWM) 0.72 0.73 0.73 0.42 0.82
Position specific scoring matrix (PSSM) 0.66 0.68 0.67 0.32 0.73
BLOSUM62 (B62) 0.61 0.58 0.59 0.15 0.60