Fig. 4
From: DrugEx v3: scaffold-constrained drug design with graph transformer-based reinforcement learning
The chemical space of generated molecules by the Graph Transformer. Shown are the molecules generated by the models pre-training on the ChEMBL set (A, C and E) and fine-tuning on the LIGAND set (B, D and F). Chemical space was represented by either logP ~ MW (A and B) and the first two components from a PCA on PhysChem descriptors (C, D) and t-SNE on ECFP6 fingerprints (E and F)