Estimation of acute oral toxicity in rat using local lazy learning

Table 1 Performance of four LLL models using different similarity metrics on the test set using reference set I (Group I) versus the best model of the reference

Models		“Set_3874”		“Set_2896”		“Set_2583”		“Set_743”
		R²	MAE	R²	MAE	R²	MAE	R²	MAE
LLR	ECFP4	0.347	0.622	0.415	0.589	0.437	0.580	0.647	0.474
	FCFP4	0.333	0.629	0.433	0.589	0.460	0.575	0.649	0.477
	MACCS	0.317	0.640	0.386	0.608	0.407	0.599	0.644	0.506
	DES	0.255	0.667	0.348	0.608	0.381	0.588	0.578	0.444
	LLR_consensus	0.434	0.535	0.511	0.508	0.534	0.499	0.738	0.401
SA	ECFP4	0.390	0.563	0.464	0.535	0.486	0.526	0.688	0.427
	FCFP4	0.363	0.581	0.469	0.543	0.494	0.527	0.674	0.441
	MACCS	0.379	0.573	0.452	0.546	0.467	0.542	0.679	0.452
	DES	0.329	0.591	0.419	0.539	0.448	0.523	0.627	0.407
	SA_consensus	0.450	0.528	0.527	0.493	0.544	0.490	0.746	0.394
SR	ECFP4	0.406	0.555	0.483	0.524	0.508	0.514	0.719	0.408
	FCFP4	0.376	0.575	0.485	0.536	0.513	0.519	0.701	0.426
	MACCS	0.383	0.570	0.457	0.543	0.473	0.539	0.690	0.445
	DES	0.329	0.590	0.419	0.539	0.448	0.523	0.627	0.407
	SR_consensus	0.457	0.515	0.536	0.489	0.555	0.485	0.761	0.384
GP	ECFP4	0.413	0.550	0.491	0.519	0.514	0.510	0.719	0.406
	FCFP4	0.367	0.583	0.477	0.543	0.504	0.527	0.689	0.437
	MACCS	0.360	0.586	0.443	0.556	0.461	0.551	0.692	0.451
	DES	0.329	0.586	0.421	0.534	0.450	0.518	0.627	0.412
	GP_consensus	0.460	0.512	0.545	0.483	0.565	0.477	0.771	0.371
Final_consensus		0.466	0.510	0.545	0.483	0.565	0.478	0.769	0.374
Reference	Individual^a	n/a^b	n/a^b	0.41	0.55	0.41	0.56	0.70	0.41
	Consensus	n/a	n/a	0.42	0.52	0.48	0.51	0.71	0.39

^aFor each coverage set, the best result among all individual models reported by Zhu et al. [7] was shown.
^bThe reference did not report the prediction results of all compounds in the test set.

ISSN: 1758-2946