Extended Data Fig. 6: Group performance evaluations.

a, Group composite Z-scores plotted by metric category. The nine teams with highest Coordinate-only composite Z-score rankings are shown, sorted left to right. The plot illustrates that by group/method, Coordinate-only scores are poorly corelated with Fit-to-Map and Comparison-to-Reference scores. In contrast, a modest correlation is observed between Fit-to-Map and Comparison-to-Reference scores. b, Averaged model composite Z-scores plotted for ab initio modeling groups at higher resolution (T1 at 1.8 Å, T2 at 2.3 Å) and lower resolution (T3 at 3.1 Å, T4 at 2.9 Å). In each case 6 groups produced very good models (Z ≥ 0.3; green pins), though not the same set. Runner-up clusters (−0.3 ≤ Z < 0.3) are shown with gold pins. Individual scores and order shift with alternate choices of evaluation metrics and weights, but the clusters at each resolution level are stable. Composite Z-scores were calculated as described in Methods.