Unraveling Protein Mysteries: HPC-Powered Exploration of Mechanostability

Protein mechanostability is crucial in numerous biological processes, yet it remains understudied. To bridge this knowledge gap, we developed an innovative modeling technique using network analysis, machine learning, molecular dynamics, and force spectroscopy measurements. This combination has been applied successfully to various systems like bacterial adhesins, cellulosomes, and viruses. Our investigation of SARS-CoV-2 spike interaction with human ACE2 protein provides a prime example of its effectiveness. Through our machine learning pipeline, we efficiently analyzed thousands of molecular dynamics simulations, pinpointing key interactions between the spike and ACE2. Our model protein, based on the 2003 SARS-CoV-1 spike with eight single-point mutations, replicated the mechanical stability of the SARS-CoV-2 spike. Our work delivers valuable insights into the mechanostability of macromolecular complexes, potentially influencing protein design and therapeutic development. In my presentation, I will discuss the specific ways in which we are combining these techniques, and provide compelling new examples that highlight their impressive capabilities.