Skip to main content
. Author manuscript; available in PMC: 2017 Sep 1.
Published in final edited form as: Proteins. 2015 Aug 14;84(Suppl 1):314–322. doi: 10.1002/prot.24862

Table 1.

Improvements over starting models for human submissions. Average improvements are given using four different measures, and the sum of Z-scores relative to other group predictions are in parentheses. Summed Z-scores are calculated using -2.0 for the minimum value, and positive values are better for all measures. Analysis with minimum Z-score = 0.0 is provided in Supplementary Table S1.


Target category Num. Targets GDT-HA (%) Sphere Grinder (%) Cα RMSD (Å) MolProbity
iGDT-HA >= 501) 242) Model1 +1.0 (+15.1) +3.8 (+23.3) -0.50 (+26.4) -1.1 (+19.7)
Top5 +3.9 (+35.0) +7.2 (+46.5) -0.84 (+45.4) -1.4 (+27.3)

iGDT-HA < 503) 12 Model1 -3.9 (-11.6) -1.1 (+2.3) +1.31 (-6.6) -1.3 (+10.0)
Top5 +2.0 (+6.4) +4.3 (+21.4) +0.27 (+1.5) -1.6 (+15.0)

Overall 36 Model1 -0.6 (+3.5) +2.2 (+25.6) +0.10 (+19.8) -1.2 (+29.7)
Top5 +3.3 (+41.4) +6.3 (+67.9) -0.47 (+46.9) -1.5 (+42.3)

1)

Targets with starting GDT-HA higher than or equal to 50.0, to which high-resolution strategy is applied.

2)

TR769 is not submitted as the target protein is engineered from Baker group.

3)

Targets with starting GDT-HA lower than 50.0, to which low-resolution strategy is applied.