A method for probabilistic mapping between protein structure and function taxonomies through cross training

BMC Structural Biology

Table 3 Performance evaluation for some structural classes

Class-Id	FP Rate	Precision	Recall	F-measure
α &β.NADP-bind Rossmann fold	0.008	0.25	0.259	0.254
Immunoglobulin β sandwich	0.018	0.296	0.282	0.289
α.globin like	0.005	0.649	0.61	0.629
Small proteins	0.063	0.245	0.216	0.213
Peptides	0.017	0.242	0.18	0.207

Performance evaluation while trying to classify structural classes using functional classes as the features. The low F-measure values indicate that though structure is dependent on function of the protein, the present features are incapable in distinguishing them completely. F-measure was calculated as the harmonic mean between Recall and Precision. Presented table lists classes which showed significant F-measure values. Class-name refers to superfamily in SCOP database. Highly significant F-measure values are shown in bold.

ISSN: 1472-6807