Table 2.

Statistics for the training datasets created by only using annotations with manual and experimental evidence codes and the training datasets created by using annotations with all evidence codes.

Training Dataset Statistics
	Annotation Count	Annotations with only manual experimental evidence codes			Annotations with all evidence codes
		# of available levels	# of GO terms	# of annotations	# of available levels	# of GO terms	# of annotations
Molecular Function	≥30	9	838	281 125	11	2 776	6 451 530
	≥100	9	605	272 235	10	1 598	6 386 105
	≥200	9	395	257 404	10	1 174	6 326 109
	≥300	8	226	233 476	9	942	6 269 643
	≥400	8	165	218 591	9	809	6 223 762
	≥500	8	142	210 790	9	698	6 173 867
Biological Process	≥30	10	4 215	1 433 220	12	8 404	16 537 812
	≥100	10	2 993	1 386 588	12	4 768	16 335 538
	≥200	9	1 782	1 302 577	11	3 299	16 129 271
	≥300	9	1 059	1 199 604	10	2 631	15 965 583
	≥400	8	743	1 123 037	9	2 233	15 828 012
	≥500	8	603	1 075 353	9	1 978	15 713 431
Cellular Comp.	≥30	7	606	340 995	8	1 268	4 167 000
	≥100	6	460	335 445	8	750	4 138 327
	≥200	6	324	325 687	7	549	4 110 383
	≥300	6	206	309 390	6	442	4 083 834
	≥400	6	155	296 929	6	377	4 061 654
	≥ 500	5	118	283 616	6	335	4 043 150