Table 1. Seven benchmark datasets used to train and test our predictor.

Dataset	Number of proteins
	all-α	all-β	α/β	α+β	total
1189	223	294	334	241	1092
D640	138	154	177	171	640
25PDB	443	443	346	441	1673
Z277	70	61	81	65	277
Z498	107	126	136	129	498
D1185	251	258	199	477	1185
D8244	1744	1929	2357	2214	8244