Fig. 9
From: SMILES-based deep generative scaffold decorator for de-novo drug design
Plots describing the resulting 4,167,207 scaffold-decorations tuples obtained from slicing 827,098 drug-like molecules from ChEMBL (see “Methods”). a The number of decorations per scaffold in the dataset ordered by the scaffolds that have the most decorations to the scaffold that has the least (notice that the y-axis is in logarithmic scale). b Histogram of the number of attachment points of the entire set. c The number of times each decoration appears in the dataset ordered from left to right (notice that the y-axis is in logarithmic scale); d Histogram of the number of RECAP bonds per molecule in the training set