Figure 8.

Detailed comparison of the top five groups by torsion and Cartesian metrics. Group names are abbreviated as follows: B‐A, Baker‐Autorefine; B, Baker; F, Feiglab; S, Seok; S‐S, Seok‐Server; NP, Naïve predictor. For each plot, dark blue bars are for residues where the mean backbone torsion angle error in the starting model was less than 30°, while red bars are for the remaining residues. A‐C. Average change from the starting model for (A) backbone torsions, (B) sidechain torsions, or (C) Cα positions. D‐F. Average residual error compared to the target for (D) backbone torsions, (E) sidechain torsions, or (F) Cα positions. Dashed lines indicate the thresholds for improvement over the naïve predictor. RMS_CA values were calculated after alignment of all Cα atoms in the models