Table 3.

The prediction performance of the proposed method under the independent dataset test.

Sequence segmentation	Dataset	Training mode	Acc	Sp	Sn	MCC
Continuous-TSSS	S. cerevisiae ($S_1$)	Default mode	0.9559	0.9697	0.9429	0.9122
		Embedding training mode	0.9412	0.9697	0.9143	0.8840
		Two channel mode	0.9559	1.0000	0.9143	0.9155
	S. pombe ($S_2$)	Default mode	0.8116	0.8788	0.7500	0.6315
		Embedding training mode	0.7826	0.8182	0.7500	0.5682
		Two channel mode	0.8116	0.9091	0.7222	0.6389
	K. lactis ($S_3$)	Default mode	0.9000	0.8462	0.9412	0.7964
		Embedding training mode	0.8667	0.8462	0.8824	0.7285
		Two channel mode	0.8333	0.8182	0.8421	0.6495
	P. pastoris ($S_4$)	Default mode	0.9500	0.9444	0.9583	0.8971
		Embedding training mode	0.9167	0.9167	0.9167	0.8281
		Two channel mode	0.9032	0.8947	0.9167	0.8009
Skip-TSSS	S. cerevisiae ($S_1$)	Default mode	0.9901	0.9794	1.0000	0.9804
		Embedding training mode	0.9901	0.9897	0.9906	0.9803
		Two channel mode	0.9951	0.9897	1.0000	0.9902
	S. pombe ($S_2$)	Default mode	0.7282	0.7353	0.7212	0.4564
		Embedding training mode	0.7379	0.6176	0.8558	0.4880
		Two channel mode	0.7573	0.7843	0.7308	0.5157
	K. lactis ($S_3$)	Default mode	0.9213	0.8667	0.9773	0.8482
		Embedding training mode	0.9213	0.8667	0.9773	0.8482
		Two channel mode	0.8539	0.9111	0.7955	0.7120
	P. pastoris ($S_4$)	Default mode	0.9454	0.9318	0.9579	0.8907
		Embedding training mode	0.9727	0.9659	0.9789	0.9453
		Two channel mode	0.9617	0.9432	0.9789	0.9238