Table 1.

Statistical parameters of quantitative structure–activity relationship (QSAR) models for di-, tri- and tetrapeptides using integrated descriptor sets.

BOSS a	Variable Number	Name of Group	Statistical Parameters b
			A	R2	Q2	RMSECV	RMSE
No	174	Dipeptides	4.000	0.948	0.874	0.222	0.142
Yes	174	Dipeptides	2.000 ± 0.604	0.950 ± 0.002	0.941 ± 0.001	0.152 ± 0.001	0.139 ± 0.002
No	261	Tripeptides	3.000	0.760	0.521	0.407	0.289
Yes	261	Tripeptides	2.000 ± 0.450	0.770 ± 0.006	0.742 ± 0.004	0.299 ± 0.002	0.282 ± 0.004
No	361	Tetrapeptides	6.000	0.965	0.682	0.429	0.143
Yes	361	Tetrapeptides	6.000 ± 1.222	0.972 ± 0.002	0.956 ± 0.002	0.160 ± 0.004	0.127 ± 0.004

^a ‘Yes/No’ indicates the model was built with/without BOSS (bootstrapping soft shrinkage) variable selection process, respectively; ^b A: the number of principle components in PLS regression; R²: the coefficient of determination; Q²: the cross-validated R²; RMSECV: the root mean square error cross validation; RMSE: the root mean square error.