Predicting Natural Evolution in the RBD Region of the Spike Glycoprotein of SARS-CoV-2 by Machine Learning

Area of Science:

• Virology
• Computational Biology
• Biotechnology

Background:

• Predicting viral mutations is critical for vaccine development and understanding evolution.
• Machine learning (ML) offers powerful tools for analyzing protein mutations and guiding directed evolution.

Purpose of the Study:

• To employ ML for predicting SARS-CoV-2 variants within the Spike glycoprotein's receptor-binding domain (RBD).
• To forecast potential future variants and assess their binding capacities.
• To validate ML regression methods and compare different predictive strategies.

Main Methods:

• Utilized PyPEF software for mutation analysis of the SARS-CoV-2 Spike RBD.
• Performed molecular dynamics simulations on predicted prospective variants.
• Developed a timestamping algorithm and auxiliary programs for efficient large-scale data processing.

Main Results:

• Predicted over 48,960,000 SARS-CoV-2 variants.
• Identified prospective variant ISOY8P5O2 with an 8.25% increase in binding energy (-110.306 kcal/mol) compared to the original strain.
• Forecasted potential emergence of variant ISOY2P5O1 around November 17, 2023 (±22 days).
• Confirmed structural significance of mutation sites using ML models.

Conclusions:

• ML effectively predicts viral evolution and protein mutation effects.
• The study provides a framework for forecasting future virus variants and informing vaccine design.
• Advanced data processing techniques enhance ML capabilities in evolutionary biology.