. 2014 Sep 6;11(18):47–58. doi: 10.1016/j.csbj.2014.09.001

Table 2.

ROC analyses of the effectiveness of three measures of protein similarity in classifying proteins in CATH folds.

	ROC area^a
Set	FASTA on native sequences	STRUCTAL on 3D sequence traces	STRUCTAL on 3D structures
Set1	0.56	0.61	0.94
Set2	0.56	0.60	0.92
Set3	0.56	0.60	0.92
Set4	0.56	0.60	0.92
Set5	0.57	0.60	0.92
Set6	0.56	0.60	0.92
Set7	0.56	0.61	0.92
Set8	0.56	0.60	0.92
Set9	0.56	0.60	0.92
Set10	0.56	0.61	0.92
CATH605	0.53	0.59	0.91

Area under the ROC curve, with 0.5 corresponding to the score of a random classifier, and 1.0 corresponding to the score of a perfect classifier.