Skip to main content

Table 5 Average per-target correlation and loss on Stage 1 and Stage 2 for DeepQA and its training features on CASP11. The significance between DeepQA and individual feature was assessed by Wilcoxon signed ranked sum paired t-test*, and its P-value was included to represent the improvement of DeepQA against its input features

From: DeepQA: improving the estimation of single protein model quality with deep belief networks

QA methods Corr. onstage 1/P-value Loss on stage 1/P-value Corr. on stage 2/P-value Loss on stage 2/P-value
DeepQA 0.64/- 0.09/- 0.42/- 0.06/-
Dope 0.54/1.77E-06 0.11/0.0421 0.30/4.63E-10 0.08/2.76E-01
EC score 0.37/4.29E-11 0.18/5.71E-07 0.02/3.23E-14 0.14/2.08E-10
GOAP score 0.54/2.74E-05 0.13/0.0016 0.31/5.07E-07 0.07/1.06E-01
ModelEvaluator score 0.56/0.0001 0.10/0.2160 0.28/1.87E-09 0.08/1.99E-02
OPUS score 0.43/2.14E-11 0.12/0.0588 0.30/4.53E-09 0.08/3.54E-01
Qprob score 0.63/0.8080 0.09/0.9382 0.38/8.63E-03 0.06/7.12E-01
RWplus score 0.54/4.80E-06 0.14/0.0009 0.30/9.41E-09 0.08/4.49E-02
SP score 0.47/3.07E-10 0.14/0.0067 0.26/6.17E-10 0.10/1.10E-05
SU score 0.50/3.78E-09 0.18/4.94E-07 0.19/6.34E-11 0.11/3.95E-07
Mean 0.52/0.09 0.13/0.14 0.27/0.00 0.09/0.17
  1. * The Wilcoxon signed ranked sum paired t-test was performed on the correlation and loss of targets between each feature against DeepQA