Table 1.

Training and test dataset compositions, strict (with non-strict in parentheses).

	sequences	PPIIH regions	PPIIH residues	non-PPIIH residues	PPIIH residues
total dataset	9333 (10 211)	15 112 (25 755)	51 337 (90 249)	2 133 861 (2 237 843)	2.3% (3.9%)
training dataset	8387 (9169)	13 645 (23 245)	46 382 (81 440)	1 880 279 (2 019 181)	2.4% (3.9%)
rest dataset	945 (1040)	1465 (2504)	4948 (8789)	201 969 (218 201)	2.4% (3.9%)