Abstract
While a plethora of different protein–ligand docking protocols have been developed over the past twenty years, their performances greatly depend on the provided input protein–ligand pair. In this study, we developed a machine-learning model that uses a combination of convolutional and fully connected neural networks for the task of predicting the performance of several popular docking protocols given a protein structure and a small compound. We also rigorously evaluated the performance of our model using a widely available database of protein–ligand complexes and different types of data splits. We further open-source all code related to this study so that potential users can make informed selections on which protocol is best suited for their particular protein–ligand pair.
Highlights
Molecular docking is nowadays a common approach in a computational drug discovery pipeline [1,2]: knowing a good approximation to the crystal pose of a ligand can provide medicinal chemists with new ideas for lead optimization that could potentially accelerate structure-based drug design
In the context of molecular docking, Deep Learning (DL) approaches have been investigated to replace classical scoring functions, showing moderate success [18,19], but still far behind the accuracy provided by standard docking procedures. Due to this fact, in this study, we explored the potential of DL approaches to both select the best possible docking protocol given a protein–ligand pair and to provide insight into which protein–ligand pairs will result in a better pose given a docking protocol
We prepared the protein–ligand refined set of the PDBbind database [21] (v.2017) according to the workflow previously described in the DockBench suite
Summary
Molecular docking is nowadays a common approach in a computational drug discovery pipeline [1,2]: knowing a good approximation to the crystal pose of a ligand can provide medicinal chemists with new ideas for lead optimization that could potentially accelerate structure-based drug design. A docking protocol can be described as the combination of a search algorithm that samples the conformational space of a ligand within a binding site and a scoring function, which quantitatively evaluates the accuracy of such poses. While in many cases the conformational search operated by docking protocols is effective in producing the correct pose for a ligand (i.e., the crystallographic pose is generally reproduced within reasonable accuracy), scoring functions often fail in ranking them (i.e., the crystallographic pose often is usually not the one with the best score) [3]. The aforementioned platform presents a benchmark of different docking protocols in a self-docking routine, whose goal is to reproduce the pose of a ligand with a known co-crystal: the ability of each protocol in producing the crystallographic pose being measured in terms of their Root Mean Square Deviation (RMSD)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
