Skip to main content

Table 5 Five-fold cross-validation results of the models trained with various features for classifying between 94 carbonylated and 412 non-carbonylated proline residues

From: Investigation and identification of protein carbonylation sites based on position-specific amino acid composition and physicochemical features

Classifier Training features Sensitivity Specificity Accuracy MCC
SVM AA 0.638 0.655 0.652 0.233
AAC 0.713 0.716 0.715 0.347
AAPC 0.646 0.728 0.713 0.309
PWM 0.745 0.733 0.735 0.388
PSSM 0.670 0.709 0.702 0.307
ASA 0.585 0.607 0.603 0.151
AAindex 0.702 0.752 0.743 0.375
J48 DT AA 0.617 0.607 0.609 0.176
AAC 0.638 0.631 0.632 0.212
AAPC 0.638 0.636 0.636 0.216
PWM 0.660 0.680 0.676 0.271
PSSM 0.670 0.709 0.702 0.307
ASA 0.574 0.583 0.581 0.123
AAindex 0.649 0.709 0.698 0.290
RF AA 0.628 0.660 0.654 0.229
AAC 0.723 0.716 0.717 0.355
AAPC 0.646 0.728 0.713 0.309
PWM 0.734 0.733 0.733 0.380
PSSM 0.660 0.704 0.696 0.294
ASA 0.585 0.607 0.603 0.151
AAindex 0.734 0.743 0.741 0.390
  1. The numbers makred with italicized font are the highest values in four measurements