Table 4.

Datasets in use.

	Name	Date of version update	Number of molecules in the original set	Number of molecules after database preparation procedures	Number of molecules in Bitter Domain	Ref
Positive set	BitterDB	05/2015	687	645	632	3
	Additional bitter	Nr	74	60	59	30
Negative set	Non-bitter in-house	Nr	1,573	1,531	1,451
	Sweet	Nr	414	414	336	30
	Tasteless	Nr	135	133	130	30
Validation sets	UNIMI	Nr		56	56
	Bitter New	Nr		29	23
	Phyto. Dictionary	Nr		89	65	45
Perspective prediction	DrugBank approved	4/2014	1,621	1,553	1,375	46
	FooDB	1/2016	24,399	20,661	13,588
	ChEBI	4/2015	44,651	37,033	27,015	29
	Natural Products from ZINC15	08/2016	38,469	28,217	27,474	50
	Sigma Ingredients Flavors & Fragrances	06/2016	1047	279	264

Nr: not relevant.