Machine learning-guided rational engineering of ACE2-derived peptides for broad-spectrum neutralization of SARS-CoV-2 variants

= 0.6958), estimated binding potentials for 54 synthetically designed peptide variants. The best lead, M1 (HAHTFLETFNYEAQTLSYE), exhibited improved docking energies (-257.35 kcal/mol for Omicron and -234.24 kcal/mol for wild-type) and improved free-binding energies (-35.67 kcal/mol and -21.63 kcal/mol, respectively), relative to the native peptide, with binding energies comparable to or exceeding reported RBD-hACE2 interaction benchmarks (- 13 to - 18 kcal/mol). Molecular dynamics analyses further confirmed enhanced conformational stability, sustained hydrogen bonding, and favourable energetic landscapes for the M1-RBD complexes. Collectively, this study demonstrates that integrating energetic decomposition with ML-driven peptide optimization provides a scalable and mechanistically informed strategy for developing peptide-based inhibitors against rapidly evolving viral pathogens.