Table 2 Confirmed hit (i.e., binder) count and hit rate from different DEL + ML combinations
From: Evaluation of DNA encoded library and machine learning model combinations for hit discovery
(a) Confirmed hit count and hit rate per DEL | |||
DEL | Number of compounds selected for experimental validation | Number of compounds identified as confirmed binders | Hit Rate |
MilliporeSigma (MS10M) | 237 | 23 | 10% |
HitGen (HG1B) | 283 | 43 | 15% |
DOS-DEL (DD11M) | 288 | 14 | 5% |
(b) Confirmed hit count and hit rate per ML | |||
ML | Number of compounds selected for experimental validation | Number of compounds identified as confirmed binders | Hit Rate |
Multilayer-Perceptron (MLP) | 217 | 24 | 11% |
Support Vector Machine (SVM) | 149 | 7 | 5% |
Random Forest (RF) | 73 | 8 | 11% |
Extra Gradient Boosting (XGB) | 163 | 9 | 6% |
Graphical Neural Network (ChemProp) | 206 | 32 | 16% |
