Examining the diversity of large collections of building blocks in 3D
We demonstrate that it is computationally feasible to cluster very large data sets of compounds according to electrostatic and shape properties. The result is an extremely information-rich fragment library.