Protein structural class prediction based on an improved statistical strategy

BMC Bioinformatics

Table 9 the 2858 dataset with cross validation test*

Class	Alpha	Beta	Alpha/beta	Alph+beta	total
DH hit number	386	415	363	193	1357
DE hit number	411	421	383	207	1422
CC hit number	562	447	334	147	1490
Class number	625	688	783	762	2858
DH accuracy(%)	61.76	60.32	46.36	25.33	47.48
DE accuracy(%)	65.76	61.19	48.91	27.17	49.76
CC accuracy(%)	89.92	64.97	42.71	19.29	52.13

*This 2858 dataset is totally different from table 6. The two datasets were obtained from the random separation of the 5714 dataset. We used the dataset in table 6 as training samples, and the sequences in table 7 were test samples. This cross validation method can be considered more reliable than jack-knife test.

ISSN: 1471-2105