Exploring the ability of the MD+FoldX method to predict SARS-CoV-2 antibody escape mutations using large-scale data

Area of Science:

• Immunology
• Computational Biology
• Virology

Background:

• Antibody escape mutations challenge vaccine and therapy effectiveness.
• Predicting these mutations computationally is crucial for early threat detection.
• Lack of large-scale experimental data hinders validation of computational methods.

Purpose of the Study:

• Evaluate the MD+FoldX molecular modeling method for predicting antibody escape mutations.
• Utilize a large deep mutational scanning dataset for validation.
• Focus on the SARS-CoV-2 receptor binding domain.

Main Methods:

• Applied the MD+FoldX molecular modeling method.
• Leveraged a deep mutational scanning dataset.
• Analyzed the SARS-CoV-2 receptor binding domain.

Main Results:

• Observed a positive correlation between predicted and experimental data.
• Reduced predicted binding affinity moderately correlated with higher experimental escape fractions.
• Tailored affinity cutoffs improved prediction performance over a general approach.
• Achieved >50% precision in 70% of tested systems.
• Successfully identified mutations in variants of concern and interest.

Conclusions:

• MD+FoldX shows potential for predicting antibody escape mutations.
• Customized affinity cutoffs enhance predictive accuracy.
• Further development of accurate and fast binding affinity prediction methods is needed.
• Challenges remain in comparing computational and experimental results.