In his latest contribution, Peter Ertl analyses the ChEMBL database to identify those ring systems that are most likely to be bioactive (< 10 µM against any target apart from hERG or CYP) alongside the background of other ring systems. When the ring systems are projected into two dimensions (via a clever adaptation of PCA that prevents outliers from dominating the distribution and which employs Scaffold Keys as descriptors), regions of ring space that are more likely to be bioactive than others are identified. The bioactivity is subdivided according to the target type (GPCR, kinase, protease, other enzyme, ion channel, nuclear receptor, transporter, epigenetic, other). The analysis reveals areas of ring space that are enriched in certain types of activity with unexplored rings in those regions looking ripe for exploitation but likely needing some synthetic tricks to be developed in order to break in. The results are available as a mind-blowing online tool!
Magic Rings: Navigation in the Ring Chemical Space Guided by the Bioactive Rings J. Chem. Inf. Model.