Table 1.

Comparison of the developed model to existing approaches.

Test dataset	Training dataset size	Model	ROC AUC	MCC
TS092	2517	PepNN-Struct	0.855	0.409
		PepNN-Struct on AlphaFold models	0.850	0.409
		PepNN-Seq	0.781	0.272
	–	AlphaFold-Multimer	–	0.605
	–	AlphaFold-Gap	–	0.518
	475	PBRpredict-flexible16	0.587	0.084
		PBRpredict-moderate16	0.569	0.071
		PBRpredict-strict16	0.543	0.049
TS251	251	PepNN-Struct	0.833	0.366
		PepNN-Seq	0.769	0.277
		Interpep11	0.793	–
	–	AlphaFold-Multimer	–	0.566
	–	AlphaFold-Gap	–	0.470
TS639	640	PepNN-Struct	0.868	0.352
		PepNN-Seq	0.795	0.246
		PepBind12	0.767	0.348
	–	AlphaFold-Multimer	–	0.450
	–	AlphaFold-Gap	–	0.432
TS125	956	PepNN-Struct	0.885	0.390
		PepNN-Seq	0.794	0.259
		BiteNetpp17	0.882	0.435
	640	PepBind12	0.793	0.372
	1156	SPRINT-Str26	0.780	0.290
	1199	SPRINT-Seq13	0.680	0.200
	1004	Visual15	0.730	0.170
	–	AlphaFold-Multimer	–	0.576
	–	AlphaFold-Gap	–	0.448

Bolded values indicate the highest metric for each dataset.