A New Approach to Design Virtual Combinatorial Library with Genetic Algorithm Based on 3D Grid Property

Combinatorial chemistry is a type of synthetic strategy which systematically connects a set of “building blocks” of varying structures to yield a large array of diverse molecular entities. Since the molecular diversity of combinatorial library used for the two stages of drug discovery process which are exploring lead compounds and performing structural modification on lead compounds is different, choosing appropriate fragments as the building blocks becomes a critical step for combinatorial chemistry. The number of available chemical fragments is often so great that it is impossible to use all of the fragments in one synthetic experiment. An effective method is much needed to select building blocks from the enormous chemical fragments to synthesize a combinatorial library with desired molecular diversity. In this article, we proposed a novel idea to define molecular diversity based on three-dimensional (3D) grid force field properties including electric potential and steric potential parameters and demonstrated how to apply genetic algorithm to choose a subset of fragment for the construction of a virtual combinatorial library. In the first example, we tried to construct a virtual library for benzodiazepine derivatives with maximum molecular diversity. In the second example, a virtual library is constructed for structural modification of (−)-huperzine A based on the 3D grid properties. By analyzing fragments frequency in the target library, we can understand which fragments are appropriate as building blocks. Furthermore, we can predict the biological activity of library compounds with 3D-QSAR model or by evaluating the binding energy of library compounds with the target receptors sites.