Skip to main content

Table 3 The performance of dipeptide composition based SVM models, evaluated using ten-fold cross validation on training data (80%) and independent validation on independent data (20%) on BalanceEval (BalanceFix & BalanceVar) dataset

From: Identification of B-cell epitopes in an antigen for inducing specific class of antibodies

Dataset   Mode Data size ACC MCC AUC
BalanceVar IgG Training 6063 70.88 0.42 0.76
Evaluation 1519 68.24 0.37 0.74
IgE Training 1873 80.53 0.61 0.87
Evaluation 468 81.49 0.63 0.88
IgA Training 322 69.60 0.39 0.75
Evaluation 80 74.69 0.49 0.79
BalanceFix IgG Training 4893 70.87 0.42 0.76
Evaluation 1223 71.67 0.43 0.78
IgE Training 1524 85.04 0.70 0.90
Evaluation 381 80.97 0.62 0.86
IgA Training 213 73 0.46 0.80
Evaluation 54 66.67 0.33 0.72