Drivers of adaptive evolution during chronic SARS-CoV-2 infections

In rare cases of chronic SARS-CoV-2 infection, dramatic evolution is observed, with mutations reminiscent of those observed in variants of concern (VOCs). We set out to consolidate results from sequencing data from a total of twenty-seven case reports, including six sequenced herein. First, we find that the overall pattern of mutations found across all chronic patients is extremely similar to the pattern found in VOCs. Surprisingly however, a subset of mutations, usually associated with high transmissibility, that have recurrent signal along the SARS-CoV-2 phylogeny are mostly absent from the chronic patients set of mutations. We suggest that epistasis may dictate when a variant formed in chronic infection and has the potential to spread onwards. Next, we went on to search for predictors of antibody-evasion mutations in our set of chronic patients. In the next stage of the research, we are continuing to probe the evolution of the virus during chronic infections by using the entire publicly available database of over 13 million SARS-CoV-2 sequences in order to identify sequences that look like they derive from chronic infections. This will be done based on their phylogenetic location in the global tree and the type of mutations that were accumulated in the viral genome, using both logistic regression and machine learning methods. This challenging process, along with the sequencing of new samples, will allow us to expand the sample of chronic infections in order to obtain a much broader view of what might lead to the creation of a new VOC.