13th Community Wide Experiment on the
Critical Assessment of Techniques for Protein Structure Prediction
Menu
TS Analysis (data assisted targets) : Group performance based on combined z-scores analysis
Results Home Table Browser Estimate of Model Accuracy Results RR Assessment Results
The cummulative z-scores in this table are calculated according to the following procedure (example for the "first" models):
1. Calculate zscores from the raw scores for all "first" models (corresponding values from the main result table);
2. Remove outliers - models with zscores below the tolerance threshold (set to -2.0);
3. Recalculate zscores on the reduced dataset;
4. Assign z-scores below the penalty threshold (either -2.0 or 0.0) to the value of this threshold.
    • Analysis on the models designated as "1"
    • Analysis on the models with the best scores
    • [Ss] SAXS targets
    • [Xx] X-link targets
    • [Nn] NMR targets
    • [Aa] SANS targets
The ranking of the groups is based on the analysis of zscores for GDT_TS.
    #     GR
    code		     GR
    name		     Domains Count     SUM Zscore
    (>-2.0)		     Rank SUM Zscore
    (>-2.0)		     AVG Zscore
    (>-2.0)		     Rank AVG Zscore
    (>-2.0)		     (>0.0)     Rank SUM Zscore
    (>0.0)		     AVG Zscore
    (>0.0)		     Rank AVG Zscore
    (>0.0)
1 196 Grudinin 4 5.7375 1 0.4344 1 3.1526 1 0.7882 1
2 122 Forbidden 4 5.1095 3 0.2774 3 2.2480 2 0.5620 2
3 288 UNRES 4 5.1564 2 0.2891 2 2.0845 3 0.5211 3
4 208 KIAS-Gdansk 4 4.5936 4 0.1484 4 1.8539 4 0.4635 4
5 492 wf-BAKER-UNRES 4 1.8513 5 -0.5372 5 0.0000 5 0.0000 5
6 329 D-Haven 4 1.5516 6 -0.6121 6 0.0000 5 0.0000 5
Protein Structure Prediction Center
Sponsored by the US National Institute of General Medical Sciences (NIH/NIGMS)
Please address any questions or queries to:
© 2007-2018, University of California, Davis