Viral burden phenotyping

With human and viral reads assigned to each droplet, we calculated viral burden for each droplet as a function of the total reads mapped to that droplet to control for differences in read depth between droplets, resulting in normalized counts of viral reads per droplet. To get burden per LCL phenotypes, we used Demuxlet assignments of droplets to LCLs to create a distribution of viral burden for each LCL. We used the mean of each distribution as the phenotype for GWAS.