16th Community Wide Experiment on the
Critical Assessment of Techniques for Protein Structure Prediction
`
TS Analysis : Z-score based relative group performance
Results Home Table Browser
  GDT_TS   Assessor's formula

    Models:

    • Ranking on the models designated as "1"
    • Ranking on the models with the best scores

    Groups:

    • All groups on 'all groups' targets
    • Server groups on 'all groups' + 'server only' targets

    Formula and Domains:

      The ranking of groups is based on the analysis of zscores for GDT_TS.
    • easy
    • medium
    • hard
    #     GR
    code
    GR
    name
    Domains Count     SUM Zscore
    (>-2.0)
    Rank SUM Zscore
    (>-2.0)
    AVG Zscore
    (>-2.0)
    Rank AVG Zscore
    (>-2.0)
    SUM Zscore
    (>0.0)
    Rank SUM Zscore
    (>0.0)
    AVG Zscore
    (>0.0)
    Rank AVG Zscore
    (>0.0)
1 163 MultiFOLD2 8 6.1737 1 0.7717 1 6.7069 1 0.8384 2
2 456 Yang-Multimer 8 4.0649 2 0.5081 3 6.3674 2 0.7959 3
3 145 colabfold_baseline 7 2.9593 3 0.7085 2 5.9823 3 0.8546 1
4 269 CSSB_server 7 0.5265 9 0.3609 4 4.9253 4 0.7036 4
5 052 Yang-Server 8 1.9972 5 0.2497 8 4.7697 5 0.5962 5
6 079 MRAFold 8 0.5708 8 0.0714 12 4.3121 6 0.5390 7
7 388 DeepFold-server 8 2.1541 4 0.2693 7 4.3005 7 0.5376 8
8 208 falcon2 7 0.4563 10 0.3509 5 4.1701 8 0.5957 6
9 312 GuijunLab-Assembly 8 -0.8085 15 -0.1011 18 3.7765 9 0.4721 9
10 148 Guijunlab-Complex 8 0.4225 11 0.0528 14 3.6550 10 0.4569 11
11 198 colabfold 7 0.2040 12 0.3149 6 3.2377 11 0.4625 10
12 425 MULTICOM_GATE 8 1.7180 6 0.2147 10 3.2098 12 0.4012 13
13 319 MULTICOM_LLM 8 -0.7338 14 -0.0917 17 3.1688 13 0.3961 14
14 375 milliseconds 8 -1.8992 20 -0.2374 23 3.0360 14 0.3795 15
15 331 MULTICOM_AI 8 -1.1453 17 -0.1432 20 2.8718 15 0.3590 16
16 267 kiharalab_server 8 0.8152 7 0.1019 11 2.5326 16 0.3166 17
17 028 NKRNA-s 5 -5.7681 26 0.0464 15 2.2523 17 0.4505 12
18 450 OpenComplex_Server 8 -4.2028 23 -0.5253 28 2.2410 18 0.2801 18
19 147 Zheng-Multimer 8 -1.5305 19 -0.1913 22 2.2354 19 0.2794 19
20 019 Zheng-Server 8 -0.9128 16 -0.1141 19 2.2328 20 0.2791 20
21 110 MIEnsembles-Server 8 -1.3649 18 -0.1706 21 2.2001 21 0.2750 21
22 314 GuijunLab-PAthreader 8 -0.3934 13 -0.0492 16 2.0076 22 0.2510 22
23 122 MQA_server 7 -4.9942 24 -0.4277 27 1.5826 23 0.2261 24
24 284 Unicorn 8 -5.5172 25 -0.6897 29 1.4026 24 0.1753 26
25 475 ptq 8 -2.5547 22 -0.3193 26 1.3931 25 0.1741 27
26 304 AF3-server 8 -2.3763 21 -0.2970 25 1.2572 26 0.1572 28
27 423 ShanghaiTech-server 5 -7.3786 28 -0.2757 24 1.1157 27 0.2231 25
28 075 GHZ-ISM 8 -5.8768 27 -0.7346 30 1.0430 28 0.1304 29
29 219 XGroup-server 2 -11.5177 29 0.2411 9 0.4823 29 0.2411 23
30 132 profold2 1 -13.9465 30 0.0535 13 0.0535 30 0.0535 30
31 361 Cerebra_server 7 -14.5569 31 -1.7938 31 0.0000 31 0.0000 31
The cummulative z-scores in this table are calculated according to the following procedure (example for the "first" models):
1. Calculate z-scores 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 z-scores on the reduced dataset;
4. Assign z-scores below the penalty threshold (either -2.0 or 0.0) to the value of this threshold.
Protein Structure Prediction Center
Sponsored by the US National Institute of General Medical Sciences (NIH/NIGMS)
Please address any questions or queries to:
© 2007-2026, University of California, Davis
Terms of Use