Table 2.

Performance comparison on structure-based RPI369, RPI2241 and RPI1807

Dataset	Method	Accuracy	Sensitivity	Specificity	Precision	MCC	AUC
RPI2241	IPMiner	0.824	0.833	0.812	0.836	0.650	0.906
	SDA-RF	0.648	0.653	0.630	0.665	0.296	0.687
	SDA-FT-RF	0.783	0.890	0.645	0.920	0.592	0.898
	RPISeq-RF	0.646	0.652	0.630	0.663	0.293	0.690
	lncPro	0.654	0.659	0.640	0.669	0.310	0.722
RPI369	IPMiner	0.752	0.735	0.791	0.713	0.507	0.773
	SDA-RF	0.707	0.699	0.727	0.689	0.416	0.754
	SDA-FT-RF	0.693	0.664	0.784	0.602	0.396	0.728
	RPISeq-RF	0.704	0.705	0.702	0.707	0.409	0.767
	lncPro	0.704	0.708	0.696	0.713	0.409	0.740
RPI1807	IPMiner	0.986	0.982	0.993	0.978	0.972	0.998
	SDA-RF	0.972	0.970	0.981	0.962	0.944	0.995
	SDA-FT-RF	0.972	0.955	0.997	0.940	0.944	0.995
	RPISeq-RF	0.973	0.968	0.984	0.960	0.946	0.996
	lncPro	0.969	0.965	0.981	0.955	0.938	0.994

The positive pairs are all from original papers. The negative pairs for RPI1807 is from original paper

The boldface indicates this measure performance is the best among the compared methods for individual dataset