Improving accuracy of protein-protein interaction prediction by considering the converse problem for sequence representation

Table 2 The performance of four symmetric representing schemes for protein pairs

Organism	Methods	Benchmark negatives					Random negatives
		AUC	Acc	Sn	Sp	Pre	AUC	Acc	Sn	Sp	Pre
E. coli	AG-CTF	0.996	0.968	0.997	0.941	0.893	0.886	0.797	0.794	0.799	0.798
	SM-CTF	0.988	0.948	0.985	0.929	0.879	0.876	0.788	0.785	0.789	0.789
	Sker-CTF	0.998	0.947	0.995	0.940	0.839	0.880	0.795	0.791	0.798	0.797
	Dist-CTF	0.955	0.892	0.842	0.899	0.786	0.769	0.702	0.704	0.699	0.701
S. cerevisiae	AG-CTF	0.991	0.968	0.991	0.959	0.786	0.948	0.880	0.879	0.928	0.919
	SM-CTF	0.990	0.964	0.984	0.958	0.766	0.939	0.868	0.837	0.899	0.893
	Sker-CTF	0.985	0.909	0.978	0.900	0.564	0.929	0.867	0.818	0.919	0.911
	Dist-CTF	0.946	0.891	0.826	0.900	0.523	0.849	0.788	0.764	0.799	0.792

Cutoff for each method was set according to the maximal F-measure statistic which is a community-standard procedure. Acc: accuracy; Sn: sensitivity; Sp: Specificity; Pre: precision.

ISSN: 1471-2105