Table 7.

Analysis of substances from the NICEATM set that were predicted incorrectly by all three in silico models, the justification for the TIMES-SS and Derek predictions are given, VEGA does not give a mechanistic description. The authors’ proposed rationales for the incorrect predictions are also given. Mechanistic domains – Michael Acceptor (MA), Schiff Base former (SB), Acyl transfer agent, SN2 etc. make reference to the reaction pathways described in Aptula and Roberts [33]

Structure	DSSTox Substance Identifier (DTXSID)	CAS	Experimental sensitisation outcome	TIMES-SS	Derek Nexus	Author proposed rationales/comments
	DTXSID60697617	190965-45-8	LLNA-NS	Aromatic carbonyl compounds	Resorcinol or precursor	SB would be more plausible if COAr was ortho to OH was an aldehyde. Resorcinol formation is deactivated by the COAr which is electron withdrawing and therefore deactivated resorcinol’s oxidation.
	DTXSID90469546	313680-92-1	LLNA-NS	Epoxides, Aziridines and Sulfuranes	Epoxide	Epoxide is probably insufficiently reaction (see [36]).
		13504-15-9	LLNA-NS	alpha, beta-Carbonyl compounds with polarised double bonds	1,2-Dicarbonyl compound or precursor	This metabolic activation is probably not significant in mouse skin vs guinea pig or man (see 50-24-8).
	DTXSID40284367	6973-09-7	LLNA-NS	Metabolite: Quinone methide(s)/imines, Quinoide oxime structure, Nitroquinones,	Aromatic primary or secondary amine	Oxidation is deactivated by SO2Me.
	DTXSID60337127	87578-93-6	LLNA-NS	alpha, beta-Carbonyl compounds with polarised double bonds	alpha, beta-Unsaturated ketone or precursor	Low reactivity as a methacrylate.
	DTXSID70345475	57514-21-3	LLNA-NS	Metabolites: Hydroperoxides	Aromatic primary or secondary amine	Oxidation of the aromatic amine would be deactivated by OMe and the fused pyridine ring.
	DTXSID9020112	1912-24-9	LLNA-NS	Activated aryl and heteroaryl compounds	Activated N-heterocycle	The negative experimental outcome is surprising. Would expect this substance to be reactive as a SNAr. Hydrolysis might account for the NS outcome reported.
	DTXSID9020112	50-23-7	LLNA-NS	Ketones	1,2-Dicarbonyl compound or precursor	Metabolic activation in mouse skin is probably not significant to lead to a sensitising outcome. (see 50-24-8).
	DTXSID50469840	444731-74-2	LLNA-NS	Activated aryl and heteroaryl compounds	Activated N-heterocycle	Surprising outcome. Would expect this substance to react as a SNAr electrophile, hydrolysis could be a competing factor.
	DTXSID70469403	444731-75-3	LLNA-NS	Activated aryl and heteroaryl compounds	Activated N-heterocycle	Surprising outcome. Would expect this substance to react as a SNAr electrophile, hydrolysis could be a competing factor.
	DTXSID40456769	201677-59-0	LLNA-NS	Epoxides, Aziridines and Sulfuranes	Alkyl halide	OH relative to the Br leaving group has a deactivating effect. Not sufficiently reactive to act as a SN2 electrophile.
	DTXSID10620233	414909-98-1	LLNA-NS	Carbamates	alpha, beta-Unsaturated ketone or precursor	Poor acyl transfer agent, weak SN2 and deactivated as MA by beta Nitrogen.
	DTXSID70725319	305366-94-3	LLNA-S	No Alerts	No Alerts	Right hand side benzene ring has 2 ortho N substituents. Hydrolysis of the amides in the 7 membered ring would give rise to a ortho-di(aryl)amino benzene which is oxidisable to a quinone imine that could act as a MA.
		168151-92-6	LLNA-S	No Alerts	No Alerts	Anionic with surfactant features, likely false positive (FP) result.
		2051-59-4	LLNA-S	No Alerts	No Alerts	MA – lactone; phenyl ester.
	DTXSID7026285	120-14-9	LLNA-S	No Alerts	No Alerts	Aromatic aldehyde SB former that happens to be sufficiently reactive to sensitise.