Skip to main content
. Author manuscript; available in PMC: 2015 Sep 1.
Published in final edited form as: Proteins. 2014 Jul 8;82(9):1850–1868. doi: 10.1002/prot.24538

Table III.

Comparison of best refinement structures submitted in CASP10 by Foldit-derived branches to the average GDT_TS value calculated over all predictions submitted by all groups.

Prediction
Target
FOLDIT
(068)
Void
Crushers
(165)
Anthropic
Dreams
(085)
Contenders
(341)
wfFUIK
(149)
wfFUGT
(260)
Average of All CASP
Predictions
(± 1 Standard Deviation)
TR644 78.55 71.99 79.61 81.92 82.45 83.33 76.57±11.47
TR655 65.86 67.57 64.00 69.29 69.00 64.57 65.08±3.74
TR661 64.32 68.65 65.68 NA 65.81 NA 74.11±7.21
TR662 81.00 83.67 82.00 83.33 84.00 83.33 79.29±9.40
TR663 69.08 77.30 76.48 NA 54.93 54.93 60.22±12.10
TR671 61.08 64.20 66.19 51.70 59.66 59.66 49.31±11.54
TR674 77.65 77.84 78.98 78.60 82.01 NA 81.76±5.08
TR679 67.08 69.72 69.97 72.24 69.85 NA 68.42±6.51
TR681 76.57 74.22 74.48 73.30 77.22 NA 70.32±14.63
TR688 69.19 74.32 72.57 NA 71.76 NA 74.18±5.37
TR689 81.08 85.05 84.93 81.42 83.88 NA 83.01±7.29
TR696 59.75 64.75 63.75 61.00 65.25 67.75 63.01±9.90
TR698 63.23 64.58 64.92 61.34 64.71 65.13 63.12±3.44
TR705 64.84 64.06 69.27 63.80 64.84 70.05 60.39±9.66
TR708 78.57 82.65 81.50 78.32 85.71 81.00 82.44±3.54
TR710 80.28 77.83 74.87 74.23 78.09 76.16 72.33±3.74
TR712 81.18 86.83 85.22 82.26 88.04 86.42 86.65±5.90
TR722 47.24 59.05 54.33 60.83 65.94 43.11 52.92±7.47
TR723 83.40 79.01 83.02 86.45 86.83 82.82 82.57±4.10
TR747 86.11 83.89 81.11 84.72 83.89 79.72 79.99±5.09
TR750 67.44 66.35 67.17 68.68 72.67 71.57 72.88±5.76
TR752 86.49 88.85 89.02 88.34 88.51 88.51 86.75±3.12
TR754 73.90 73.90 75.73 60.66 71.69 56.25 70.26±7.54
# Wins 2 3 3 2 7 4 2

Besides the average GDT_TS value, the corresponding standard deviations are provided for each target. GDT_TS values were tabulated using the prediction center’s official results on the CASP10 webpage2. NA indicates that a structure was not submitted and evaluated for this target. Group numbers are denoted in parentheses. The best prediction for each target among these methods is bolded. Foldit-based human groups Void Crushers and Anthropic Dreams were able to refine and select the best structure among the cohort three times each, whereas FOLDIT was able to select it for two targets. In these two cases though, Foldit’s selection produced the top improvement in GDT_TS of any method in CASP10. These teams were a collaboration of citizen scientists. The WeFold branches wfFUIK selected the best structure seven times and wfFUGT four times indicating the enhanced performance of using these synergistic branches utilizing the structures refined by the citizen scientists followed by subsequent clustering, further refinement, and selection.