Abstract The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) main protease (Mpro) is a cysteine protease that is essential for viral replication. A virtual alanine scan for all residues of SARS-CoV-2 Mpro was performed in this study, which resulted in a total of 289 artificial mutants. Threshold values for root mean square deviation (RMSD)–based classification to extract drug resistance candidates were investigated. Known resistance and nonresistance mutations were defined using a publicly available database, and statistical metrics were utilized to evaluate the validity of the thresholds. We further examined whether changing the thresholds would help us correctly identify mutants that are already known to cause drug resistance. By using the best criterion determined by the F1 score, 33 mutants were identified as potential drug resistance candidates. Under the most refined prediction conditions, 23 mutants—which include five that have been experimentally reported to confer drug resistance—were identified as potential resistance-related mutations. Molecular dynamics simulations revealed that certain distant mutations, like P252A and T304A, can alter ligand configurations. Virtual alanine scan combined with optimized RMSD-based criteria can provide a practical framework for predicting resistance-related mutations, even for residues located far from the catalytic dyad.
Mizuno et al. (Sat,) studied this question.