Property Biased-Diversity Guided Explorations of Chemical Spaces

Abstract

Discovering functionally useful structures and materials by exploring the vastness of chemical space is an exciting undertaking. If done efficiently, one can discover structures that can have therapeutic value (such as drug like organic molecules) or technological value (such as organic light emitting diodes). While mining of chemical space has the potential to generate libraries of functional structures and materials, one can also easily be lost in its vastness (~1060 theoretically possible small organic molecule ~500 Da molecular weight or less). We have developed a strategy that allows efficient explorations of vast chemical spaces to generate libraries of functional organic molecules. The method, at its core, applies physical properties and structural diversity biased sampling of chemical space to search for new structures. We demonstrate the soundness and efficiency of this approach by searching through the known and enumerated databases to discover diverse organic molecules with optimum electronic and biophysical properties, and we also compare it to various existing approaches used for molecular search and property optimization. We also show a practical application of this approach by designing libraries of chromophores that emit light in the blue region of the spectrum as well as potential leads for protein and RNA binding.