Enhancing the understandings on SARS-CoV-2 main protease (M^pro) mutants from molecular dynamics and machine learning

While star drugs like Paxlovid have shown remarkable performance in combating SARS-CoV-2, we still face serious challenges such as viral mutants and resistance. In this study, we employ a computational framework combining molecular dynamics (MD) simulations, enhanced sampling techniques, and machine learning (ML) approaches to systematically investigate the molecular mechanisms underlying drug resistance in SARS-CoV-2 main protease (M^pro) mutants. Specifically, based on the accuracy of the analytical structures and the advantages of MD simulation, we deeply analyze the influence of mutants on drug resistance and its intrinsic function from the dynamic dimension. The relevant data for M^pro with different states are compared and analyzed to consolidate the understanding of mutant effect. Through the free energy perturbation method, the absolute binding free energy diagrams of M^pro mutants and Nirmatrelvir are provided, which is meaningful to the design, comparison and optimization of the new-generation inhibitors. The interaction pattern between M^pro mutants and substrate is unraveled with the AlphaFold3 model, effectively filling the deficiency of experiments. Moreover, ML model is used to explore the differentiated synergetic pathways with the important dual mutants. The critical sites in the protein network are provided, which emphasizes on the importance and urgency of in-depth research on similar systems.