KnowTox: pipeline and case study for confident prediction of potential toxic effects of compounds in early phases of development

Table 4 Information on KnowTox-AA and other CP methods using the random forest ML algorithm to predict androgen receptor antagonism

Method	Data source: actives/inactives	CP aggregation method^a	Descriptors
KnowTox-AA	ToxCast: 868/5842	ACP	Morgan+MACCS (\(+\hbox {physchem}\))
eMolTox [35]^b	\(\hbox {Literature}\):^d (1) 532/6207 (2) 406/6256	ACP	\(\hbox {Morgan + physchem}\)
Norinder et al. [33]^c	Jensen et al. [42]: 293/637	CCP	(1) Dragon (2) Signatures (3) Physchem

^aACP aggregated conformal predictor, CCP cross-conformal predictor [48]
^bTwo models ((1), (2)) fitted on two different AA datasets
^cThree models ((1), (2), (3)) with different fingerprints trained on one AA dataset
^dData for a total of 174 CP models originated from ChEMBL, Pubchem, Toxnet, eChemPortal databases and literature [35]

ISSN: 1758-2946