Inferring selection effects in SARS-CoV-2 with Bayesian Viral Allele Selection

Area of Science:

• Virology
• Genomics
• Evolutionary Biology

Background:

• Millions of SARS-CoV-2 genomes offer insights into viral evolution.
• Analyzing genomic data to understand viral selection is crucial for vaccine and treatment development but presents analytical challenges.

Purpose of the Study:

• To develop a scalable probabilistic method for inferring genetic determinants of viral fitness and lineage growth rates.
• To identify mutations influencing SARS-CoV-2 fitness, including those dependent on vaccination status and epistatic interactions.

Main Methods:

• Development of Bayesian Viral Allele Selection (BVAS), a probabilistic method combining Bayesian variable selection and diffusion approximation.
• Application of BVAS to analyze 6.9 million SARS-CoV-2 genomes.
• Extension of the BVAS model to detect fitness changes related to vaccination and mutation interactions (epistasis).

Main Results:

• BVAS accurately infers genetic determinants of viral fitness through simulations.
• Identification of numerous fitness-increasing mutations in SARS-CoV-2 Spike, Nucleocapsid, and non-structural proteins.
• Discovery that mutations affecting fitness, particularly at the N501 residue in the Spike protein, show strong dependence on vaccination status and epistatic interactions.

Conclusions:

• BVAS provides a robust framework for identifying fitness-associated mutations in large-scale genomic surveillance data.
• The study highlights the significant role of the N501 residue in SARS-CoV-2 evolution, influenced by host immunity and genetic interactions.
• The findings contribute to understanding SARS-CoV-2 evolution and inform the design of more effective vaccines and therapeutics.