Using big sequencing data to identify chronic SARS-Coronavirus-2 infections

Area of Science:

• Virology
• Genomics
• Immunology

Background:

• Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) evolution is marked by emerging divergent variants.
• Chronic infections in immunocompromised individuals are hypothesized to be a source of these variants, but data is limited.

Purpose of the Study:

• To identify potential chronic SARS-CoV-2 infections and associated mutations using genomic data.
• To infer mutations characteristic of chronic infections and assess their evolutionary impact.
• To develop a method for identifying chronic infections and predicting future viral evolution.

Main Methods:

• Phylogenetic analysis of millions of SARS-CoV-2 genomes to identify chronic-like clades (≥21 days, identical metadata).
• Application of language models (LM) to infer chronic-associated mutations, particularly in the spike receptor-binding domain (RBD).
• Comparison of mutation frequencies and evolutionary success between chronic and global transmission chains.

Main Results:

• Identified 271 chronic-like clades exhibiting characteristics of prolonged infections.
• Chronic-associated mutations included high-fitness, immune-evasive RBD mutations, with some unique to chronic settings.
• High-fitness RBD mutations were 10-20 times more probable in chronic infections than global chains.
• The majority of BA.1/BA.2 RBD mutations from chronic clades predicted global success.
• LM inferred hundreds of additional chronic-like clades without metadata.

Conclusions:

• Chronic SARS-CoV-2 infections are a significant reservoir for the emergence of high-fitness, immune-evasive variants.
• Genomic surveillance and computational approaches can identify chronic infections and predict viral evolution.
• Understanding chronic infection dynamics is crucial for anticipating future SARS-CoV-2 evolutionary trajectories.