Table 1. The two benchmark API datasets that are used for building two different classification models.

Source	Number of positive pairs	Number of negative pairs	Description
[18]	580	1740	Training data for API classifiers
	145	435	Validation data for API classifiers
[21]	157	493	Training data for API classifiers
	56	56	Validation data for API classifiers and Apta-MCTS

Note that we obtained two pre-trained API classifiers for the aptamer generative model, Apta-MCTS, using the benchmarks.