Table 1. A comparison of the proposed predictor with the existing methods based on the 10-fold cross-validation on the same 250 carbonylated proteins.

Predictor	Metrics and graph	Type of carbonylation
		K	P	R	T
PTMPreda	Acc (%)d	88.59	82.93	86.64	88.39
CarSpredb		87.22	82.93	86.22	86.61
iCar-PseCpc		84.43	86.79	84.23	86.17
PTMPreda	MCCd	0.1892	0.2573	0.1878	0.2186
CarSpredb		0.2268	0.2331	0.2245	0.2040
iCar-PseCpc		0.5906	0.6006	0.6076	0.6185
PTMPreda	Sn (%)d	23.45	21.43	20.02	22.38
CarSpredb		23.17	25.34	25.47	21.39
iCar-PseCpc		45.18	48.20	46.67	50.68
PTMPreda	Sp (%)d	92.99	93.20	90.99	91.36
CarSpredb		92.43	93.28	93.39	93.42
iCar-PseCpc		99.25	98.54	99.57	98.58
PTMPreda	AUCe	0.6858	0.6903	0.5981	0.6563
CarSpredb		0.6849	0.7163	0.7158	0.7134
iCar-PseCpc		0.8728	0.8484	0.8668	0.8603

^aThe predictor developed in [33], where ξ = 13; i.e. the sample length is 27.

^bThe predictor developed in [32], where the sample length was not fixed.

^cThe predictor proposed in this paper.

^dSee Eq.9 for the definition of metrics.

^eThe area under the curve of Figure.2; the greater the AUC value is, the better the corresponding predictor will be [52, 53].