| Drug development is a time-consuming, expensive area of research that requires the collaboration of different fields of expertise. In this context, computer-aided methods have the potential to critically shorten the times and monetary expense required for preclinical development. Very early in a drug discovery effort, docking programs are used to virtually screen large libraries of compounds searching for a lead compound that can be developed into an efficient and safe drug. One of the critical components of a docking program is the one providing an estimate of the binding affinity of the formed complex: the scoring function. The central goal of this work was to develop methods to predict the binding affinity of potential drugs for application in virtual screening campaigns. Towards this goal, we first examined the current status of docking programs and scoring functions applied to a metalloprotein target relevant for medicinal chemistry, Golgi α-mannosidase II. This triggered a more in-depth analysis of scoring functions, which led us to assemble a set of protein complexes and analyze the performance of different available scoring functions in flexible and solvated proteins. We then set out to develop a scoring function based on molecular mechanical force fields and additional parameters accounting for entropic costs and solvation. We also reached into the development of molecules binding to nucleic acids by developing a hybrid docking/molecular dynamics method to study transition metal complexes binding to G-quadruplexes. Our docking program, FITTED, was prepared for the application to virtual screening campaigns with the implementation of the developed scoring functions and the development of additional tools to manipulate, select and prepare large libraries of ligands. With this goal, we developed a module for FITTED, termed SMART, to prepare ligands prior to the docking. The core of SMART was used to build two other modules, REACTOR and SELECT, that have applications in the preparation and clustering, respectively, of virtual libraries of ligands. |
