Machine Learning-Enhanced Molecular Dynamics Analysis of Residue Variant Effects on Human ACE2 and Furin Protein Interactions with SARS-CoV-2 Spike

The interaction between the Spike glycoprotein and its host cell receptors ACE2 and Furin determines the replication rates of SARS-CoV-2. Given the emergence of variants, the exact mechanisms of these interactions are not yet fully understood. This study utilizes molecular modeling, protein interaction, and molecular dynamics (MD) simulations to investigate the stability and interactions of the Wild-type (WT) and variants of ACE2-Spike-Furin complexes. MD simulations produce a vast amount of data that can be challenging to interpret. Machine learning (ML) techniques were used to reduce the complexity of the data and assess the impact of residues in these interactions. The aim is to better understand the impact of residues on these interactions, with a specific focus on the interaction region between Spike and Furin. The results demonstrate that the modeled complexes remained stable and provided valuable data for ML models. The Random Forest model proves particularly effective in identifying significant residues in each variant, revealing 500,000 residues pairs involved in the Spike-Furin interaction. These results highlight the importance of not only the active interaction sites but also the subtle interactions that are distant from the active site. By utilizing ML, critical residues can be identified, potentially leading to targets for therapeutic interventions.