Skip to main content

Table 5 Target Family Performance Comparison

From: Industry-scale application and evaluation of deep learning for drug target prediction

 ExCAPE-MLAstraZenecaJanssen
Target familyTargets w./sz.p-valueTarget w./sz.p-valueTargets w./sz.p-value
Oxidoreductase25/372.17e057/183.91e−0116/323.77e−02
Transferase141/1581.58e4877/1433.64e−0760/1537.38e−02
Hydrolase63/926.24e1221/476.96e−0236/779.91e03
Lyase4/82.59e−010/0 4/82.59e−01
Isomerase4/61.00e−010/11.00e + 004/61.00e−01
GPCR Fam. A76/943.78e2134/705.87e0366/931.39e13
GPCR Fam. B3/52.10e−010/51.00e + 002/55.39e-01
GPCR Fam. C3/52.10e−011/58.68e-015/54.12e-03
Nuclear Hormone Receptor14/208.79e049/177.55e−0213/191.87e03
Reader2/77.37e−011/13.33e−012/21.11e−01
Eraser7/98.28e−035/74.53e-024/71.73e-01
Writer3/33.70e−021/13.33e−010/11.00e + 00
Ligand-gated3/63.20e−011/48.02e−012/66.49e−01
Voltage-gated6/121.78e−013/96.23e−016/121.78e−01
Primary active3/41.11e−010/21.00e + 002/44.07e−01
Electrochem.7/101.97e−022/88.05e−013/85.32e−01
Overall364/4761.26e82162/3382.02e08225/4385.95e15
  1. Number of targets won (w.) by DNNs from a target family, size of target family (sz.) and p-values of binomial tests for each target family class, with the null hypothesis that the probability of being the best method for a certain target is less than 1/3 for DNNs when compared to XGB and MF
  2. p-values below the significance threshold of 0.01 are in italics