Identification of TBK1 inhibitors against breast cancer using a computational approach supported by machine learning

Abstract

Introduction: The cytosolic Ser/Thr kinase TBK1 is of utmost importance in facilitating signals that facilitate tumor migration and growth. TBK1-related signaling plays important role in tumor progression, and there is need to work on new methods and workflows to identify new molecules for potential treatments for TBK1-affecting oncologies such as breast cancer. Methods: Here, we propose the machine learning assisted computational drug discovery approach to identify TBK1 inhibitors. Through our computational ML-integrated approach, we identified four novel inhibitors that could be used as new hit molecules for TBK1 inhibition. Results and Discussion: All these four molecules displayed solvent based free energy values of -48.78, -47.56, -46.78 and -45.47 Kcal/mol and glide docking score of -10.4, -9.84, -10.03, -10.06 Kcal/mol respectively. The molecules displayed highly stable RMSD plots, hydrogen bond patterns and MMPBSA score close to or higher than BX795 molecule. In future, all these compounds can be further refined or validated by in vitro as well as in vivo activity. Also, we have found two novel groups that have the potential to be utilized in a fragment-based design strategy for the discovery and development of novel inhibitors targeting TBK1. Our method for identifying small molecule inhibitors can be used to make fundamental advances in drug design methods for the TBK1 protein which will further help to reduce breast cancer incidence.

Keywords: TBK1 inhibitor; binding energy; breast cancer; hydrogen bonding; machine learning; molecular dynamics.