Within-host evolution of Influenza Virus A

While the global evolutionary dynamics of influenza viruses are dominated by positive selection of antigenic variants, there is little evidence for positive selection of antigenic variants within deeply sequenced within-host populations. This discrepancy could be attributable to technical difficulties with rare variant detection, insufficient longitudinal sampling, or differences in evolutionary dynamics across scales. We examined variant detections in specimens collected longitudinally during influenza A virus (IAV) infection. We sequenced samples from individuals enrolled in FluTES, a case ascertained household study of influenza across three Influenza Seasons (2017/18-2019/20). Infected individuals were sampled daily for up to 14 days. All samples with a qPCR cycle threshold less than or equal to 30 had two technical replicates sequenced using the Illumina platform. We successfully sequenced 346 IAV positive nasal swabs (185 H3N2 and 161 H1N1) from 143 individuals. Intrahost single nucleotide variants (iSNV) were called at a 0.5% frequency threshold against a within sample consensus using iVar. We inferred loci under positive selection in two different ways: parallel evolution between individuals and allele trajectories within individuals. In the latter analysis, we estimated selection coefficients using a Wright-Fisher Approximate Bayesian Computation (WFABC) model and determined if iSNV reached consensus. We found that most within-host variants were either selectively neutral or removed by purifying selection. However, we found evidence for positive selection within the rare variant fraction that is detectable with extremely high depth of coverage, a low variant frequency threshold, and serial sampling. Nearly all sites inferred to be under positive selection laid outside of HA antigenic regions.