VINARDO2: A NEW SCORING FUNCTION FOR MOLECULAR DOCKING WITH IMPROVED VIRTUAL SCREENING

QUIROGA, RODRIGO; VILLARREAL, MARCOS A.

Salta

Congreso; PABMB-SAIB 2019; 2019

Molecular Docking is a computational method which aims to predict the optimal position and orientation of a ligand binding to a protein, as well as the strength of this interaction. Docking is a key tool in structure-based drug design. In essence, the method consists of the search for the global maximum of a mathematical function which represents the predicted affinity of the interaction between a ligand and a protein. This mathematical function, termed scoring function, is developed based on different approaches. Among them, there are semi empirical, fully empirical, knowledge and artificial-intelligence based methods. For the past few years, we have been developing an empirical scoring function based on the well-known Vina scoring function. Autodock Vina was the most cited and used docking software worldwide in 2018 (more than 1700 citations in 2018 and 8800 since it´s publication in 2010). In this work we present the newest version of our scoring function which includes a completely reworked hydrogen bond function, improved atom-typing, improved exploration of ligand conformational space, and a novel solvation term. The philosophy behind our development was to come up with coarse grained interactions, where the atomic parameters that define the interactions are picked from a set of predefined and discrete values. The performance of the Vinardo2 scoring function was evaluated in rescoring, redocking and virtual screening tasks. The results show that the new function performs significantly better at all three tasks when compared with our previous version (Vinardo) and also with the original Vina scoring function. This improved performance of Vinardo2 comes with no additional computational cost or complicated ligand preparation procedures. This makes Vinardo2 an all-around improved scoring function which can be used in direct replacement of the Vina scoring function with improved performance, especially in regards to virtual screening and computational drug discovery.