CRNPRED: highly accurate prediction of one-dimensional protein structures by large-scale critical random networks

BMC Bioinformatics

Table 2 Summary of per-residue accuracies for SS predictions.

measure	H	E	C
Q _s	82.7	69.3	84.0
$Q_{s}^{p r e}$	84.4	78.9	78.3
MC	0.754	0.674	0.645

Q_s: The number of correctly predicted residues of the SS class s = H, E, C divided by the number of residues in the class in native structures.
$Q_{s}^{p r e}$ : The number of correctly predicted residues of the SS class s = H, E, C divided by the number of residues predicted as the corresponding class.
MC: Matthews' correlation coefficient.

ISSN: 1471-2105