Table 3.

The cross-validation performances of various algorithms using the alternative secondary structure assignment scheme.

		SP175					SP175(nr)
Method	Parameter	α-helix	βD	Core β-sheet	PP-II helix	other	α-helix	βD	Core β-sheet	PP-II helix	other
SIMPLS	r	0.968	0.895	0.875	0.687	0.842	0.968	0.883	0.861	0.678	0.839
	δ	0.054	0.022	0.034	0.036	0.052	0.055	0.023	0.036	0.036	0.052
PCR	r	0.966	0.894	0.876	0.684	0.841	0.965	0.881	0.862	0.677	0.837
	δ	0.056	0.022	0.034	0.036	0.052	0.057	0.023	0.035	0.036	0.053
PLS	r	0.971	0.889	0.863	0.641	0.839	0.970	0.881	0.854	0.628	0.833
	δ	0.052	0.023	0.035	0.038	0.052	0.053	0.024	0.036	0.039	0.053
PLS-opt	r	0.971	0.868	0.867	0.702	0.835	0.969	0.856	0.846	0.696	0.830
	δ	0.053	0.025	0.035	0.035	0.053	0.054	0.026	0.037	0.035	0.054
SELMAT3	r	0.957	0.911	0.811	0.640	0.827	0.954	0.888	0.751	0.530	0.772
	δ	0.063	0.021	0.041	0.039	0.054	0.065	0.023	0.047	0.043	0.062
SELMAT1_norm	r	0.958	0.923	0.815	0.669	0.796	0.955	0.903	0.774	0.668	0.771
	δ	0.062	0.019	0.040	0.037	0.058	0.065	0.021	0.045	0.037	0.062

The (nr) tag indicates that the cross-validation was carried out under the more stringent conditions where no proteins in the training set with the same CATH homologous superfamily as that of the test protein were included. The best results for each secondary structure type for the standard and non-redundant datasets are shown in bold.