A machine learning approach for predicting methionine oxidation sites

BMC Bioinformatics

Table 1 Performance rates with three different ML models

Feature set	AUC	Accuracy	Sensitivity	Specificity	F-measure	MCC
TRAINING SET
RF
Primary (52)	1.0000	0.8233	1.0000	0.7980	0.5868	0.5756
Tertiary (24)	0.9958	0.7222	1.0000	0.6823	0.4746	0.4607
Whole (76)	1.0000	0.8476	1.0000	0.8258	0.6222	0.6107
mRMR (54)	1.0000	0.8348	1.0000	0.8111	0.6031	0.5918
SVM
Primary (52)	1.0000	0.4955	1.0000	0.4231	0.3322	0.2903
Tertiary (24)	0.9403	0.9232	0.8571	0.9327	0.7368	0.7024
Whole (76)	0.9927	0.9910	0.9592	0.9956	0.9641	0.9590
mRMR (54)	0.9952	0.9821	0.9490	0.9868	0.9300	0.9200
NN
Primary (52)	0.7148	0.6492	0.6020	0.6559	0.3010	0.1764
Tertiary (24)	0.7981	0.7273	0.7143	0.7291	0.3966	0.3132
Whole (76)	0.7827	0.6402	0.8061	0.6164	0.3599	0.2822
mRMR (54)	0.7933	0.6786	0.8061	0.6603	0.3863	0.3156
TESTING SET
RF
Primary (52)	0.7002	0.5969	0.8125	0.5664	0.3333	0.2500
Tertiary (24)	0.8014	0.6357	0.8750	0.6018	0.3733	0.3155
Whole (76)	0.8413	0.7597	0.8125	0.7522	0.4561	0.3998
mRMR (54)	0.8462	0.7597	0.7500	0.7611	0.4364	0.3668
SVM
Primary (52)	0.5603	0.4264	0.7500	0.3805	0.2449	0.0894
Tertiary (24)	0.4701	0.2791	0.6250	0.2301	0.1770	-0.1106
Whole (76)	0.6831	0.7984	0.4375	0.8496	0.3500	0.2431
mRMR (54)	0.7406	0.7907	0.4375	0.8407	0.3415	0.2320
NN
Primary (52)	0.5669	0.5504	0.4375	0.5664	0.1944	0.0026
Tertiary (24)	0.8291	0.7364	0.8125	0.7257	0.4333	0.3742
Whole (76)	0.7959	0.6589	0.7500	0.6460	0.3529	0.2661
mRMR (54)	0.8208	0.7132	0.8750	0.6903	0.4308	0.3839

ISSN: 1471-2105