Prediction of blood–brain barrier penetrating peptides based on data augmentation with Augur

Table 1 Comparison of multiple ML methods for identifying B3PPs

Evaluation strategy	ML method	Sn	Sp	ACC	MCC	AUROC
Training set	RF	0.800	0.819	0.809	0.620	0.885
	LightGBM	0.809	0.819	0.814	0.630	0.879
	LR	0.767	0.716	0.742	0.487	0.822
	SVM	0.749	0.730	0.740	0.481	0.801
	KNN	0.791	0.772	0.781	0.563	0.807
Independent set validation	RF	0.852	0.822	0.824	0.454	0.924
	LightGBM	0.852	0.848	0.848	0.489	0.922
	LR	0.833	0.760	0.767	0.375	0.879
	SVM	0.778	0.783	0.782	0.364	0.867
	KNN	0.852	0.764	0.772	0.390	0.883

ISSN: 1741-7007