Table 3.

Evaluation indicators of different deep-learning network architectures with 5-fold cross-validation, including CNN- LSTM, CNN and LSTM, respectively

Residue type	Algorithm	(%)	(%)	(%)
S/T	CNN-LSTM	80.45	79.70	81.19	0.6102	0.8871
	CNN	80.05	76.53	83.54	0.6035	0.8773
	LSTM	79.22	73.50	84.89	0.5903	0.8655
Y	CNN-LSTM	75.22	74.85	76.18	0.5183	0.8414
	CNN	76.05	76.36	76.18	0.5371	0.8055
	LSTM	66.48	68.33	65.09	0.3630	0.6739