A simple case to illustrate ChemTree

We collect 107 unique molecules with pIC50 from CSAR Protein-Ligand Information. Then, we build a tree with those molecules. The tree has three distinctive branches with Dibenzodiazepinone series (A), Dihydroindenopyrazole series (B) and Phenylpyrazinylurea series (C).

The tree has several features.

The molecule close to the center of the tree is simpler than the molecule in the terminal (figure above).

The color shows that similar molecules have similar activities. So, SAR information could be mined by groups of molecules (figure above).

The potency information is utilized to annotate ChemTree nodes. When randomly shuffle the experimental potency values, the tree does not display any interpretable SAR information (figure above). So, ChemTree is a meaningful way to organize molecules.