Selection of potential ligands for TRPM8 using deep neural networks and intermolecular docking

Abstract

Introduction. TRPM8 has been implicated in the development of bronchial hypersensitivity to cold and is considered a potential target for computer-generated drugs.Aim. Development of a strategy for the selection of ligands for TRPM8 by in silico methods.Materials and methods. Using machine learning tools based on deep neural networks and further verification by intermolecular docking, a strategy has been proposed for predicting potential ligands for TRPM8, which consists in using a neural network to screen out potential drug candidates and thereby reduce the list of candidate ligands for verification using AutoDock program, which allows assessing the affinity of a protein for a ligand by the minimum binding energy and identifying possible conformations of a ligand upon binding to certain centers (amino acid residues) of a protein. The latter were used: Y745 (tyrosine 745 is a critical center for TRPM8), R1008 (phenylalanine 1008) and L1009 (alanine 1009).Results. Of the 10 potential ligands predicted by the neural network, eight showed a high minimum binding energy and a greater number of conformations compared to the classic TRPM8 ligand, menthol, when verified by the AutoDock program. The two predicted ligands did not show the ability to interact with TRPM8, which may be due to insufficient allocated memory of the computing device for successful docking or other technical problems.Conclusion. The proposed strategy is universal; it will accelerate the search for ligands for various proteins and will facilitate the accelerated search for potential drugs by in silico methods.