Molecular Design and Virtual Screening of EZH2 Inhibitors Based on GRU Networks

EZH2 (Enhancer of Zeste Homolog 2) is a key epigenetic regulator implicated in various cancers, making the discovery of potent and selective EZH2 inhibitors a significant goal in drug development. Deep learning methods applied to chemistry can be used to accelerate the discovery of new molecules. In this paper, a platform for smiles string-based molecular design using gated recurrent neural networks (GRU) was used for the ab initio design of EZH2 small molecule inhibitors. The generated virtual library was screened by conventional machine models 、ADMET、 Lipinski and DFT. The combination of virtual screening and molecular dynamics for further simulation studies of the dominant provides guidance and theoretical basis for the generation and structural optimization of EZH2 inhibitors. This study demonstrates the potential of deep learning-based methods in efficiently generating and optimizing novel small molecule inhibitors, which provides a promising strategy for future drug discovery efforts targeting EZH2.