TABLE 2.

Performance Comparisons of DeepT3_4 with other three methods on the independent test dataset 3.

Model	TP	TN	FN	FP	ACC	F-value	Recall	PRE	MCC
					(%)	(%)	(%)	(%)
Bastion3	33	72	2	3	95.5	93.0	94.3	91.7	0.896
Bastion4	27	75	8	0	92.7	87.1	77.1	100.0	0.835
CNN-T4SE	28	75	7	0	93.6	88.9	80.0	100.0	0.855
DeepT3_4	32	72	3	3	94.5	91.4	91.4	91.4	0.874