Table 1.

Summary of datasets used in the study for the DeepChem-based method.

Dataset Type	Split	Datapoints	% Active/Toxic Compounds	Sources
Antibacterial Activity	Training	98,261	0.31	In-house + SPARK + COADD databases
	Validation	24,590	0.36
	Test	2,208	1.63
Toxicity	Training	40,566	4.78	In-house + Clintox + Tox21 databases
	Validation	10,141	4.77
	Test	12,676	4.76
XGBoost	Training	385	N/A	KEGG database
	Test	97	N/A