Table 1.

Evaluation of peptide and protein classifiers

Test	Classifier	MCC	ACC	F1-score	AUROC
Filtered	Peptide	0.74	0.87	0.87	0.92
	Protein	0.68	0.83	0.83	0.94
Unfiltered	Peptide	0.72	0.86	0.86	0.92
	Protein	0.63	0.80	0.80	0.92

Matthews correlation coefficient (MCC), accuracy (ACC), F₁-score and area under ROC curve (AUROC) of protein and peptide classifiers for blind tests are shown. MCC, ACC and F₁-score are reported at threshold score $\ge 0.9$. The filtered set has no missing values for any of the features, whereas unfiltered includes all feature data (as would be the case in a real world prediction scenario).