Skip to main content

Advertisement

Table 5 Virtual Screening results for Factor Xa at five different applicability levels

From: Estimation of the applicability domain of kernel-based machine learning models for virtual screening

Kernel ADE   Threshold AUROC BEDROC (α) Ligands Decoys
      100.00 53.6 32.2 20.0   
OAK KDE 50% 0.645 0.78 0.59 0.52 0.51 0.49 58 1020
   33% 0.65 0.78 0.68 0.62 0.61 0.57 57 661
   200 0.67 0.74 0.58 0.68 0.69 0.69 52 148
   100 0.68 0.62 0.36 0.53 0.54 0.61 40 60
  wKDE 50% 0.65 0.78 0.59 0.52 0.51 0.49 58 1020
   33% 0.66 0.78 0.65 0.59 0.58 0.55 57 661
   200 0.67 0.74 0.58 0.68 0.69 0.69 53 147
   100 0.68 0.62 0.36 0.55 0.58 0.64 42 58
  ONE 50% -15.09 0.78 0.60 0.53 0.53 0.50 58 966
   33% -13.98 0.78 0.65 0.58 0.51 0.53 58 660
   200 -10.84 0.73 0.58 0.68 0.69 0.69 56 144
   100 -7.33 0.61 0.36 0.55 0.57 0.64 45 55
FlexOAK KDE 50% 0.64 0.76 0.42 0.38 0.38 0.37 59 1019
   33% 0.64 0.75 0.43 0.41 0.41 0.39 58 660
   200 0.66 0.62 0.11 0.18 0.18 0.22 47 153
   100 0.67 0.48 0.00 0.03 0.04 0.07 43 57
  wKDE 50% 0.64 0.76 0.40 0.37 0.37 0.36 59 1019
   33% 0.65 0.74 0.40 0.39 0.38 0.37 57 661
   200 0.66 0.63 0.11 0.18 0.18 0.22 48 153
   100 0.67 0.48 0.00 0.03 0.03 0.07 43 57
  ONE 50% -15.40 0.76 0.41 0.38 0.38 0.37 59 1016
   33% -13.86 0.75 0.43 0.42 0.42 0.40 59 659
   200 -10.92 0.64 0.13 0.23 0.24 0.29 51 149
   100 -8.20 0.42 0.00 0.03 0.04 0.08 44 56
MARG KDE 50% 0.84 0.57 0.00 0.02 0.02 0.3 54 1024
   33% 0.85 0.57 0.00 0.01 0.02 0.03 44 674
   200 0.864 0.48 0.00 0.00 0.00 0.00 21 179
   100 0.866 0.56 0.00 0.00 0.00 0.00 16 84
  wKDE 50% 0.845 0.57 0.00 0.01 0.02 0.03 54 1024
   33% 0.852 0.56 0.00 0.00 0.00 0.01 43 675
   200 0.86 0.47 0.00 0.00 0.00 0.00 21 180
   100 0.87 0.55 0.00 0.00 0.00 0.00 16 84
  ONE 50% -11.30 0.56 0.01 0.02 0.02 0.03 55 1014
   33% -9.95 0.55 0.04 0.01 0.01 0.02 43 675
   200 -7.79 0.41 0.00 0.00 0.00 0.00 14 186
   100 -7.09 0.47 0.00 0.00 0.00 0.00 8 92
  1. The threshold is adjusted such that either a certain fraction (50%, 33%) of the compounds or that exactly n compounds (n = 200, 100) remain in the domain. Results, which differ significantly from random rankings (p-Value < 0.01) are shown in bold.