Skip to main content
. 2023 Feb 3;15(2):516. doi: 10.3390/pharmaceutics15020516

Figure 5.

Figure 5

Compound activity predictions for CCR2 using the sequential Keras/TensorFlow model of NN. Histogram of ligand activities as predicted by Keras/TensorFlow. Ligand activities (pChEMBL) were divided into ranges (x-axis). The fraction of the dataset that was assigned to each activity range (in %) was given in brackets. Predicted activity values fell into three categories: overpredicted, underpredicted, and predicted correctly, with (left) and without (right) inactive compounds included in the datasets.