Skip to main content

Table 2 Known drug status prediction (± standard deviation across 5 folds)

From: Empowering the discovery of novel target-disease associations via machine learning approaches in the open targets platform

Method

LogReg

RF

XGB

Train set

OT association evidence

 AUROC

0.7603 (± 0.0088)

0.8685 (± 0.0075)

0.8784 (± 0.0051)

 AUPR

0.0685 (± 0.0025)

0.2074 (± 0.0093)

0.2072 (± 0.0103)

Computed features + OT association evidence

 AUROC

0.8867 (± 0.0027)

0.9262 (± 0.0019)

0.9406 (± 0.0018)

 AUPR

0.6442 (± 0.0069)

0.7500 (± 0.0070)

0.7969 (± 0.0065)

Validation set

OT association evidence

 AUROC

0.7625 (± 0.0357)

0.8143 (± 0.0314)

0.8076 (± 0.0335)

 AUPR

0.0707 (± 0.0102)

0.0872 (± 0.0118)

0.0888 (± 0.0177)

Computed features + OT association evidence

 AUROC

0.8864 (± 0.0076)

0.9103 (± 0.0140)

0.9137 (± 0.0142)

 AUPR

0.6452 (± 0.0226)

0.7092 (± 0.0459)

0.7264 (± 0.0457)