Table 3.

Five-fold cross-validation results of the models trained with various features for classifying between 101 carbonylated and 504 non-carbonylated arginine residues

Classifier	Training features	Sensitivity	Specificity	Accuracy	MCC
SVM	AA	0.614	0.603	0.605	0.163
	AAC	0.653	0.683	0.678	0.259
	AAPC	0.663	0.687	0.683	0.270
	PWM	0.713	0.718	0.717	0.336
	PSSM	0.624	0.685	0.674	0.239
	ASA	0.594	0.599	0.598	0.145
	AAindex	0.693	0.726	0.721	0.329
J48 DT	AA	0.554	0.603	0.595	0.119
	AAC	0.594	0.683	0.668	0.214
	AAPC	0.614	0.687	0.674	0.233
	PWM	0.614	0.675	0.664	0.222
	PSSM	0.554	0.665	0.646	0.169
	ASA	0.535	0.599	0.588	0.101
	AAindex	0.646	0.690	0.683	0.259
RF	AA	0.614	0.605	0.607	0.165
	AAC	0.634	0.683	0.674	0.244
	AAPC	0.653	0.683	0.678	0.259
	PWM	0.713	0.716	0.716	0.334
	PSSM	0.624	0.685	0.674	0.239
	ASA	0.594	0.599	0.598	0.145
	AAindex	0.693	0.724	0.719	0.327

The numbers marked with italicized font are the highest values in four measurements