Table 1.

Input feature type performance comparison results.

Model & GO level	GO term id	GO description	# of annotated proteins	Predictive performance (F1-score)
				SPMap	Pseudo-amino acid composition	Conjoint triad
Model 1 (GO level: 2)	GO:0036094	small molecule binding	1 847	0.49	0.29	0.23
	GO:0003700	DNA binding transcription factor activity	1 652
	GO:0004872	receptor activity	1 332
	GO:0044877	protein-containing complex binding	1 296
	GO:0097367	carbohydrate derivative binding	1 252
Model 2 (GO level: 4)	GO:0004529	exodeoxyribonuclease activity	50	0.68	0.53	0.38
	GO:0045309	protein phosphorylated amino acid binding	50
	GO:0008395	steroid hydroxylase activity	49
	GO:0008649	rRNA methyltransferase activity	49
	GO:0015645	fatty acid ligase activity	49
Model 3 (GO level: 7)	GO:0001012	RNA polymerase II regulatory region DNA binding	818	0.74	0.53	0.47
	GO:0016887	ATPase activity	764
	GO:0046873	metal ion transmembrane transporter activity	685
	GO:0001159	core promoter proximal region DNA binding	504
	GO:0015077	monovalent inorganic cation transmembrane transporter activity	480