Conquering Chemical Space : Optimization of Docking Libraries through Interconnected Molecular Features

Abstract: Copied selected text to selection primary: The development of new pharmaceuticals is a long and ardous process that typically requires more than 10 years from target identification to approved drug. This process often relies on high throughput screening of molecular libraries. However, this is a costly and time-intensive approach and the selection of molecules to screen is not obvious, especially in relation to the size of chemical space, which has been estimated to consist of 10 60 compounds. To accelerate this exploration, molecules can be obtained from virtual chemical libraries and tested in-silico using molecular docking. Still, such methods are incapable of handling the increasingly colossal virtual libraries, currently reaching into the billions. As the libraries continue to expand, a pre-selection of compounds will be necessitated to allow accurate docking-predictions. This project aims to investigate whether the search for ligands in vast molecular libraries can be made more efficient with the aid of classifiers extended with the conformal prediction framework. This is also explored in conjunction with a fragment based approach, where information from smaller molecules are used to predict larger, lead-like molecules. The methods in this project are retrospectively tested with two clinically relevant G protein-coupled receptor targets, A 2A and D 2 . Both of these targets are involved in devastating disease, including Parkinson’s disease and cancer. The framework developed in this project has the capacity to reduce a chemical library of > 170 million tenfold, while retaining the 80 % of molecules scoring among the top 1 % of the entire library. Furthermore, it is also capable of finding known ligands. This will allow for reduction of ultra-large chemical libraries to manageable sizes, and will allow increased sampling of selected molecules. Moreover, the framework can be used as a modular extension on top of almost any classifier. The fragment-based approaches that were tested in this project performed unreliably and will be explored further.