Skip to main content

Table 2 Result for drug target protein prediction using machine learning methods

From: In silico re-identification of properties of drug target proteins

SVM Recall Precision F1
Set A, W 0.7326 0.6594 0.6941
Set A, W 0.7516 0.7422 0.7469
Set A, W+N 0.7947 0.6681 0.7259
Set A, W + N 0.8137 0.6982 0.7515
Set B, W 0.7866 0.6416 0.7067
Set B, W 0.7374 0.6496 0.6907
Set B, W+N 0.7424 0.6585 0.6979
Set B, W + N 0.8018 0.6580 0.7228
Set C, W 0.7516 0.7808 0.7659
Set C, W 0.7972 0.8003 0.7987
Set C, W+N 0.8137 0.7965 0.8050
Set C, W + N 0.8409 0.8207 0.8307
Set D, W 0.7820 0.7367 0.7587
Set D, W 0.8083 0.7588 0.7828
Set D, W+N 0.8120 0.7500 0.7798
Set D, W + N 0.8271 0.7710 0.7981
RF    
Set A, W 0.7541 0.7682 0.7605
Set A, W 0.6483 0.8130 0.7260
Set A, W+N 0.7936 0.6763 0.7299
Set A, W + N 0.8229 0.6986 0.7556
Set B, W 0.7821 0.6547 0.7124
Set B, W 0.7490 0.6493 0.6953
Set B, W+N 0.7551 0.7805 0.7677
Set B, W + N 0.8076 0.6767 0.7363
Set C, W 0.7847 0.7358 0.7589
Set C, W 0.8165 0.7960 0.8057
Set C, W+N 0.8292 0.8118 0.8200
Set C, W + N 0.8509 0.8218 0.8354
Set D, W 0.7885 0.7409 0.7636
Set D, W 0.8343 0.7564 0.7934
Set D, W+N 0.8305 0.7550 0.7908
Set D, W + N 0.8382 0.7818 0.8088
  1. Feature sets W and N represent widely used and newly proposed properties, respectively. W and N represent statistically significant widely used and newly proposed properties, respectively
  2. The underline bold numbers indicate the highest values in each evaluation