Table 11 Performance comparison results between different methods on the independent test datasets
From: Positive-unlabelled learning of glycosylation sites in the human proteome
Type | Methods | AUC | ACC | F1 |
|---|---|---|---|---|
C-linked | PA2DE (V2.0) | 0.999 | 0.983 | 0.984 |
GlycoEP | 0.546 | 0.600 | 0.647 | |
ModPred | 0.933 | 0.933 | 0.938 | |
N-linked (motif) | PA2DE (V2.0) | 0.893 | 0.815 | 0.820 |
GlycoEP | 0.697 | 0.637 | 0.638 | |
NetNGlyc | 0.638 | 0.616 | 0.627 | |
ModPred | 0.837 | 0.782 | 0.791 | |
N-linked (non-motif) | PA2DE (V2.0) | 0.872 | 0.761 | 0.758 |
GlycoEP | 0.669 | 0.648 | 0.644 | |
NetNGlyc | 0.630 | 0.716 | 0.675 | |
ModPred | 0.842 | 0.807 | 0.773 | |
O-linked (S) | PA2DE (V2.0) | 0.915 | 0.859 | 0.766 |
GlycoEP | 0.796 | 0.787 | 0.848 | |
ModPred | 0.873 | 0.770 | 0.670 | |
NetOGlyc | 0.770 | 0.662 | 0.583 | |
O-linked (T) | PA2DE (V2.0) | 0.864 | 0.793 | 0.827 |
GlycoEP | 0.739 | 0.694 | 0.747 | |
ModPred | 0.821 | 0.746 | 0.781 | |
NetOGlyc | 0.769 | 0.718 | 0.763 |